Potential of Copper- and Cerium-doped Optical Fiber ...

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TNS.2016.2606622, IEEE Transactions on Nuclear Science

Potential of Copper- and Cerium-doped Optical Fiber Materials for Proton Beam Monitoring 1

S. Girard, Senior Member IEEE, B. Capoen, H. El Hamzaoui, M. Bouazaoui, G. Bouwmans, A. Morana, D. Di Francesca, A. Boukenter, O. Duhamel, P. Paillet, Senior Member IEEE, M. Raine, Member, IEEE, M. Gaillardin, Member, M. Trinczek, Member, IEEE, C. Hoehr, E. Blackmore, Member, IEEE and Y. Ouerdane signal, and decreases the signal-to-noise ratio (SNR), especially when both spectral domains overlap. RIE can also be due to defects that are light emitting centers through Radiation-Induced Luminescence (RIL) [2] and/or to Cerenkov light emission for higher energy particles [3]. Fewer applications are affected by the RIE than by the RIA. Thus, RIE is today more exploited as an innovative way to monitor the radiation flux than considered as a limiting factor for data transmission [2,3]. Several fiber-based systems have already been developed and are used for the radiation monitoring; a review is given in [4]. Very promising results were recently obtained for passive dosimetry using the thermo-luminescence properties of optical fibers that are able to measure the equivalent deposited dose (in rad(SiO2)) for various types of particles: protons, neutrons, X-ray or γ-rays [5]. Even more interest exists for online dosimetry allowing to follow in real-time the delivered dose and allowing to control the time profile of the dose rate. Several fiber-based technologies were investigated but for medical applications (eg. Radiotherapy), scintillating optical fibers based on plastic (PMMA) materials doped with scintillators, such as Gadolinium Oxysulfide activated with Terbium [6], were developed and are now commercially-available. However, if these sensors are sensitive enough to radiations in terms of RIL or Cerenkov [7], their main drawback remains in their high sensitivity to RIA due to the degradation of the polymer host matrix under radiations that prevents their use for high dose rate/high dose applications [8]. Thereafter, their lifetime is limited to low dose levels. One of the goal of this study consists in extending the dose rate/dose (or flux/fluence) measurement using such fiber-based dosimeters by using a different fiber technology. This latter consists in a tip of scintillating SiO2-based optical material of a few mm in length coupled to passive optical fiber to transport the RIE signal to the detector. This is a first step of our approach, the future objective being to draw out the developed glasses in the form of microstructured optical fibers (similar to the pure and F-doped ones recently characterized in [9] or to the Cu-doped photonic crystal fiber (PCF) fiber studied in [10]) and to develop suitable

Abstract –We investigate the potential of innovative optical fiber bulk materials made by the sol-gel technique for realtime proton beam monitoring. Those types of glass are made of amorphous silica (a-SiO2) doped with either Copper (Cu) or Cerium (Ce) ions. These optimized materials possess very interesting light emission properties when exposed to protons. For both types of glasses, online monitoring of the strong radiation-induced luminescence (RIL) allows to monitor the time evolution of the proton flux with ms resolution and the cumulated proton fluence can be precisely deduced by integrating the RIL signal. Furthermore, we showed that both samples present optically-stimulated luminescence (OSL) that could be exploited shortly after the end of the irradiation to reconstruct the cumulated fluence too. Preliminary tests presented in this paper have been performed at TRIUMF facility with high energy (30 - 63 MeV) protons at flux and fluences representative of proton therapy treatments. Obtained results demonstrate the potential of the developed optical materials for proton beam monitoring and permits to identify the needs for future experiments on microstructured fibers elaborated from these optical materials. Index terms– dosimetry, proton therapy, optical fibers, copper, cerium, radiation-induced attenuation, radiationinduced emission, optically-stimulated luminescence, radioluminescence.

I. INTRODUCTION

A

morphous silica (a-SiO2)-based optical fibers are considered for implementation in various nuclear fields for data transfer systems, sensors or diagnostics [1]. This technology presents a much higher radiation tolerance level than those of most micro- or optoelectronic technologies. Today, specialty optical fibers for applications above the MGy (100 Mrad) dose levels are commercially-available and present limited degradation of their optical properties in severe environments [1]. Below 1MGy dose level, ionizing or non-ionizing radiations cause two main macroscopic degradation mechanisms of the a-SiO2 optical properties. Radiation Induced Attenuation (RIA) that decreases the fiber transparency is caused by the generation of lightabsorbing point defects. Radiation Induced Emission (RIE) is a parasitic effect that superimposes to the guided optical 11

Manuscript received July 8th, 2016, revised August 31th, 2016. S. Girard, A. Morana, D. Di Francesca, A. Boukenter and Y. Ouerdane are with Univ-Lyon, Laboratoire H. Curien, UMR CNRS 5516, 18 rue du Pr. Benoît Lauras 42000, Saint-Etienne, France. (Phone : +33477915812, [email protected] ) B. Capoen; H. El Hamzaoui; M. Bouazaoui, G. Bouwmans are with Univ-Lille, CNRS, UMR8523 - PhLAM – Physique des Lasers, Atomes et Molécules, CERLA/IRCICA, F-59000 Lille, France: ([email protected]) O. Duhamel, P. Paillet M. Raine, and M. Gaillardin are with CEA, DAM, DIF, F-91297 Arpajon, France (e-mail : [email protected] ). M. Trinczek, C. Hoehr and E. Blackmore are with TRIUMF, Vancouver, Canada (e-mail : [email protected])

1 0018-9499 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.


real-time spatially-resolved dosimetry system in the future. These new materials were elaborated by the sol-gel technique, allowing to incorporate inside silica bulk, Cu+ or Ce3+ ions associated with interesting RIE properties while keeping the host matrix radiation resistant to RIA (as puresilica is one of the most radiation hardened type of glass [1]). Combining these two advantages, the developed sensors could be used in a larger range of applications than the existing ones. These glasses have been previously evaluated under X-rays and ultraviolet (UV) laser light [1013]. This paper focuses on their potential for beam monitoring during protontherapy treatment. To the best of our knowledge, none work was devoted to this issue. Then, the performances of Cu- and Ce-doped materials need to be characterized under protons of energy, flux and fluences representative of such medical treatments. This paper presents the first results obtained on the bulk optical materials in an extrinsic fiber-dosimeter scheme at TRIUMF facility with experimental conditions representative of the treatments delivered at this facility.

B. RIL and OSL Experimental Setup The goal of this first study was to evaluate the potential of such glasses for dosimetry purposes by proving their capacities to exhibit RIL and OSL responses. To be promising, these induced emissions should be intense enough, located in the visible and should exhibit exploitable dose rate (RIL) and dose (RIL, OSL) dependences. The sample emission signal is guided from the bulk sensing element via one of these fibers to an air tight box containing selective 550 nm band pass filters to eliminate the stimulating laser light at 660 nm in the case of OSL experiments. It should be noted that the choice of this filter is adequate for the Cu-doped sample as it is known that the induced emission related to Cu+ ions is peaked around 540 nm [10,11,13]. More recently, we showed that, in the case of the Ce-doped glasses, the X-ray RIL related to Ce3+ ions is centered around 460 nm [13], meaning that the setup used for these experiments causes a clear underestimation of the RIL and OSL signal amplitude for Ce-doped glasses. We used for the detection a low-cost and compact photomultiplier tube (PMT, Model H9305-03, Hamamatsu). In the case of OSL experiments, the excitation signal was delivered from a fibered laser diode at 660 nm (LP660-SF60, Thorlabs). This stimulation wavelength was chosen as a compromise between a high stimulation efficiency for emitting ions in silica and the absence of photoluminescence (PL) generation.

II. TESTED OPTICAL MATERIALS & IRRADIATION CONDITIONS A. Tested Optical Fiber Materials Ionic Copper- or Cerium-doped glassy rods were prepared at PhLAM laboratory of Lille according to procedures described elsewhere [10-13]. As an example, for the Cudoped glass, a monolithic porous pure silica xerogel cylinder, synthesized through a base-catalyzed sol-gel process and using tetraethylorthosilicate as a precursor, was soaked in a solution of copper (II) hexafluoroacetylacetonate hydrate. Our previous studies [10-13] showed that, after drying the samples in air to remove the solvent, it is better to densify both Cu and Ce-doped silica xerogels in a helium gas atmosphere to promote the RIE phenomena by increasing the Cu+ and Ce3+ ions concentrations [11,12] as these ions are the ones with the stronger emission efficiency. Finally, thin transparent rods of ~0.75 (Cu rod) and ~1 mm (Ce) size diameters were drawn from these bulk samples. For RIL and opticallystimulated luminescence (OSL) measurements, ~25 mmlength pieces of these thin rods were optically polished at both ends, put in a fitted pure-silica sleeve and connected at both ends to 0.5 mm-diameter multimode pure-silica core optical fibers (PSCF).

III. EXPERIMENTAL RESULTS A. Evidence for RIL and RIA Phenomena in Both Glasses Previous experiments demonstrate the potential of these two glasses to monitor the dose rate evolution under UV light or 10 keV X-rays [10-13] through RIL. We first investigate in this section, the responses of the two sol-gel materials exposed to 63 MeV protons in a range of flux and fluences similar to that one used for some of the TRIUMF protontherapy treatments: equivalent dose of about 1 krad(Si) deposited in few tens of seconds. The whole 25 mm length of Cu- and Ce-doped thin rods are exposed to the proton beam (square beam of 5cm×5cm) in the back position of the BL2C proton line [14]. It was checked under X-rays that no RIL is observed from the pure-silica core pigtails used to connect the rods to the PMT and to the probe 660nm laser. Using the PMT, we are able to detect if the glasses exhibit RIL signal around 550 nm and to follow its amplitude kinetic with a 5 ms resolution. A few seconds after blocking the proton beam, RIL disappears and then 660 nm laser light is launched into the sample irradiated at different fluences to characterize its potential in terms of OSL under this excitation wavelength. Typical results are given in Figure 2a for the Cu-doped sample and in Figure 2b for the Cedoped sample exposed to 63 MeV protons at ~20°C, with the highest beam current of 7 nA up to a dose of 1 krad (7.53×108 p/cm²). During the exposure to 63 MeV protons at room temperature (RT), each fiber material type exhibits a strong RIL phenomena that is easily detected by our non-optimized detection chain with a good SNR. Additional tests (not reported here) confirmed that this emission directly follows the TRIUMF BL2C current time evolution with a time resolution well below the second. Observed longer time fluctuations such as the RIL increase observed during the run illustrated in Fig.2b are then more probably related to

Fig.1. Illustration of the experimental setup used to characterize the RIL and OSL response of our fiber material exposed to protons with energies from 30 to 63 MeV.



0.3 a) Cu-doped sample

B. Consecutive Runs at Fixed Proton Energies and Beam Currents Figure 3 illustrates the observed signals for both materials in a series of consecutive irradiation runs, varying the dose while keeping constant the beam current around 7nA.

PMT output signal (V)


the proton beam fluctuation (eg. online adjustment of the beam shape) than to the material response. It should be noted that today setup did not authorize a direct comparison between the proton RIL efficiencies of both materials as the rod-to-fiber coupling (illustrated in Fig.1) remains sampledependent and also, the used filter is not well-adapted to CeOSL Laser ON

63 MeV irradiation

PMT noise level

200

400

600

800

a) Cu-doped sample, 63MeV, 7nA 1krad

0.9

2krad

0.3

zone 1

-400

1000

-200



(b) Ce-doped

0.15 63 MeV irradiation

0.10

OSL Laser probe ON

PMT noise level

Time (s)

200

400

b) Ce-doped sample, 63MeV, 7nA

0.2

200

1krad

2krad

3krad OSL

OSL

OSL

0.1

0.0

0.00 100

0

Time (s)

0.20

0

OSL

0.6

Time (s)

0.05

500rad

1krad OSL

OSL

0.0

0.0 0

1.2

-400

-200

0

200

400

Time (s) Fig.3. Illustration of both the RIL and OSL phenomena measured during successive 63MeV proton irradiation runs at RT for (a) Cu-doped and (b) Ce-doped samples. The equivalent irradiation dose during each run was varied between 0.5 and 3 krad(Si) with a stable beam current set at 7nA.

Fig.2. Illustration of the RIL and OSL phenomena measured at room temperature during 63MeV proton exposure of (a) Cu-doped (b) Ce-doped samples.

doped glass characterization. However, previous works showed a very high PL quantum efficiency (QE) of 40% [11] and 33% [12] for such inorganic materials under UV laser excitation for the Cu and Ce-doped glasses, respectively. We observe a quick rise-time for the RIL at the beginning of the run and a rapid decay time of this emission at the end of the proton exposure. After opening the beam blocker the RIL signal is characterized by a first fast rise in less than 5 ms that seems followed by a slower multi-exponential increase as it was observed for X-rays [12,13]. A few seconds after the run stop, the injection of the probe laser at 660 nm provides evidence for an OSL phenomenon in both optical materials. Here again, the comparison between the performances of the two materials cannot easily be performed today but it remains possible to compare the relative amplitudes of the RIE and OSL in both materials in the same irradiation conditions. OSL response of the Cudoped samples reveals a higher relative amplitude than in Ce-doped glass. It is very important to notice that if the OSL has already been observed for Cu-doped materials, it is the first time to our knowledge that a Ce-doped material is able to exhibit both RIL and OSL signatures.

First, RIL level appears not significantly affected by the accumulated dose resulting from the successive tests and remains stable at given irradiation conditions (flux, fluence). For the Cu-doped sample, the OSL response clearly increases with the dose (from 0.5 to 2 krad). From these preliminary results, it is also observed that the currently used filters are not sufficient to fully remove the probe laser light that still contributes to the detected signal in addition to the OSL phenomenon (zone 1 in Fig.3a). This prevents us from performing a precise evaluation of the OSL response with fluence but first tendencies will be discussed later in the paper. For the Ce-doped sample, the OSL signal is lower as expected (as the filters are not centered at the maximum emission peak). Hence, its dose dependence seems unclear from these preliminary measurements. OSL signal maximum seems stable between the second and third runs, even if the dose was increased by 50% (from 2 to 3 krad). However, it should be noted that for the last run, the delay time between the end of the proton exposure and the laser injection was increased, leading additional time for the OSL emitting centers to thermally bleach. This OSL vs delay time response will be studied in the future. As a preliminary conclusion, it seems that if both



glasses exhibit strong RIL signal, the developed Cu material appears as more promising in terms of OSL response.

evolution should allow an easy and real-time monitoring of the proton flux (equivalent dose rate) with a ms resolution at a specified location point within a mono-energetic proton beam.

IV. DISCUSSION A. Equivalent dose rate evaluation through RIL We performed a systematic study of the RIL and OSL signal evolutions with the proton flux (hereafter converted into equivalent dose rate in rad/s) and the fluences (here converted into equivalent doses in rad(Si)) for each proton energy. All obtained results are then exploited to characterize the RIL amplitude dependence on the equivalent dose rate. Obtained results for both glasses are given in Figure 4, we extract the RIL mean value by averaging the PMT output signal over the whole irradiation run, as illustrated in Fig4a for three runs at different beam currents for the Cu-doped sample. The equivalent dose rate has been obtained using the conversion factor 1 rad(Si)= 7.53×105 p(63MeV)/cm² provided by TRIUMF. Fig.4b gives the dependence of the mean RIL value (expressed in V) versus the equivalent dose rate (expressed in rad/s).

For both glasses, we observe a linear trend for the RIL vs dose rate, with sensitivity coefficients of 3.65×10-2 (Copper, Pearson’s r=0.9943) and 5.10×10-2 (Cerium, Pearson’s r=0.98702). We notice a larger dispersion of the whole set of experimental points than the one at 63MeV for both types of glass, clearly larger for Ce-doped glass.

Cu-doped sample

0.5 11.6 rad/s

0.4 0.3

8.1 rad/s

0.7

0.2

PMT output voltage (V)

PM output signal (V)

0.6

B. RIL Evolution by Changing Proton Energy We also performed a preliminary investigation of the radiation response of our materials at various proton energies by using the PIF installation degrader [14]. At all the tested energies (35, 48 and 57 MeV), the samples exhibit OSL and RIL signatures similar to those shown in Fig.2 under 63 MeV protons. In figure 5, we plot the equivalent dose rate dependences of the RIL response of the Cu-doped (Fig.5a) and the Ce-doped (Fig.5b) samples using the following fluence-to-dose conversion coefficients provided by TRIUMF for silicon: • 57MeV: 1 rad(Si) = 7.02×105 p/cm² • 48MeV: 1 rad(Si) = 6.18×105 p/cm² • 35MeV: 1 rad(Si) = 4.79×105 p/cm²

2.4 rad/s

0.1 0.0 0

100

200

300

400

500

Time (s) Cu-doped sample (63MeV) Ce-doped sample (63MeV)

0.5

y=0.0392x (Pearson's r= 0.99706)

0.4

35MeV 48MeV 57MeV 63MeV

0.5 0.4

y=0.03655x

0.3 0.2 0.1 0.0 2

4

6

8

10

12

14

16

18

20

Equivalent dose I rate (rad/s)

0.3 0.2

PMT output voltage (V)

PM output signal (V)

0.6

a) Cu-doped sample 0.6

y=0.00637x (Pearson's r= 0.99979)

0.1 0.0 0

2

4

6

8

10 12 14

16

18 20

Equivalent dose rate (rad(Si)/s) Fig.4. (a) Illustration of the dose rate dependence of the measured RIL for the Cu at 3 different dose rates and the same dose of 1krad. (b) Illustration of the dose rate dependence of the measured RIL for the Cu and Ce-doped materials under exposure to 63MeV proton beam. Tests were performed at room temperature, varying the beam current from 1 to 7nA.

Both glasses exhibit a linear dose rate dependence of the RIL in the investigated ranges of proton flux corresponding to dose rate between 0 and 15 rad/s and this at all proton energies. From these curves, one could estimate the RIL sensitivity coefficient to 3.92×10-2 and 6.37×10-3 V/(rad/s) for the Cu-doped and Ce-doped rods, respectively. This linear dependence of RIL with dose rate is a crucial property for protontherapy purposes, as the recording of the RIL

0.12 b) Ce-doped sample 63MeV 48MeV 57MeV 35MeV

0.09

0.06 y=0.0051x

0.03

0.00 2

4

6

8

10

12

14

16

18

20

Equivalent dose rate (rad/s) Fig.5. Dose rate dependence of the observed RIL for the Cu (a) and Cedoped (b) materials under exposures to 35, 47, 58, 63 MeV protons. Tests were performed at RT, varying the beam current from 1 to 7nA.



C. Monitoring the Equivalent Dose through RIL The equivalent dose (in rad) could be monitored through both RIL and OSL measurements. In the next paragraphs, we first discuss the dose estimation based on the integration of the radioluminescence (iRIL) signal during the whole irradiation run and then the one based on the OSL signal integration (iOSL) after the end of the proton exposure.

D. Monitoring the Equivalent Dose through OSL: a Preliminary Analysis for Cu-doped Glass Figure 7 illustrates the dose dependence of the OSL signal for the Cu-doped sample when exposed to 63 MeV protons at beam currents from 1 to 7 nA. To build Fig.7, we calculate the integral of the area under the OSL (iOSL) peak after subtraction of the baseline related to the small parasitic signal occurring from the laser probe (zone 1 of Fig.3a). These results provide a clear evidence for a non-linear dose dependence of the iOSL up to 5 krad, where the set of results acquired after 1 krad equivalent dose highlights almost no impact of the dose rate (beam current) on the dose evaluation. Furthermore, results at other energies (not reported here) reveals that the iOSL seems dependent on the energy of protons too. Using the actual setup configuration, the OSL signal in the Ce-doped one is too noisy to authorize today an accurate analysis of its properties. Future experiments appear mandatory with filters adapted to Ce3+ optical signature (around 460 nm).

Figure 6 gives the dependence of the integrated RIL (iRIL) versus the equivalent dose as provided by TRIUMF for both Cu- and Ce-doped materials, and this for protons of energy ranging from 35 to 63 MeV and different beam currents (from 1 to 7 nA). Results illustrated in Fig.6a on the Cudoped sample reveal a quasi-linear dependence of the iRIL versus the equivalent dose. Here again is observed a slight influence of the proton energy on the iRIL value at a given dose and without significant dose rate influence. The same linearity of the iRIL versus the dose is observed for the Cedoped sample at the various energies but, as shown in Fig.6b, there exists a stronger influence of the proton energy on the iRIL, whereas no influence of dose rate (beam current) is observed in this case too.

iOSL (arb. units)

16 Cu-doped sample

Integrated RIL (arb. units)

250 a) Cu-doped sample 200 35 MeV 7 nA 35 MeV 3 nA 48.8 MeV 7 nA 48.8 MeV 3 nA 57 MeV 7 nA 57 MeV 3 nA 63 MeV 7 nA 63 MeV 6 nA 63 MeV 4 nA 63 MeV 2 nA 63 MeV 1 nA

150 100 50

1

2

3

4

5

6

10 8 6

63 MeV 7 nA 63 MeV 6 nA 63 MeV 4 nA 63 MeV 2 nA 63 MeV 1 nA

4

0 0

1

2

3

4

5

6

Equivalent dose (krad)

7

Fig.7. Evolution of the integrated optically-stimulated luminescence (iOSL) with the equivalent dose deposited by 63 meV protons, beam currents from 1 to 7 nA in the Cu-doped sample.

Equivalent Dose (krad) Integrated RIL (arb. units)

12

2

0 0

14

30 b) Ce-doped sample

C. Origin of the Proton Energy Influence on RIL Amplitude and Integral: influence of sample geometry

35 MeV 7 nA 35 MeV 4 nA 35 MeV 2 nA 48.8 MeV 7 nA 48.8 MeV 4 nA 57 MeV 7 nA 57 MeV 4 nA 63 MeV 7 nA 63 MeV 4 nA 63 MeV 2 nA

25 20 15 10

When analysing in more details the behavior of the RIL amplitude coefficients for each proton energy (Fig.5), we observe an almost perfect linear dose rate dependence at each energy and for both materials, with r-value above 0.999 for all results (except for Cu @ 63MeV where r = 0.997). All obtained coefficients are given in the inset of Fig.8 that illustrates the proton energy dependence of both glass RIL coefficient sensitivities.

5 0 0

1

2

3

4

Equivalent dose (krad) Fig.6. Evolution of the integrated radiation-induced luminescence (iRIL) with the equivalent dose deposited by protons of energy ranging from 35 to 63 MeV, beam currents from 1 to 7 nA (a) Cu-doped sample, (b) Cedoped sample.

From these preliminary measurements, the iRIL in both Cuand Ce-doped materials is a very efficient way, after calibration, to monitor simultaneously the proton flux and fluence of a mono-energetic proton beam. However working at different energies of protons implies different calibrations of the dosimetry system.


RIL sensitivity coeff.


0.08

0.06

a dependence is rarely observed for optical fibers in which ionization is mainly responsible for the observed radiationinduced optical changes such as RIA, RIE (except for high fluences of neutrons or heavy ions [17]). As an example, we recently showed that the Ge-doped fibers exhibit a thermoluminescence signature (associated with the GLPC defects) that is independent on the particle nature (neutron, proton, X-ray, γ-ray….) at various energies [5]. To investigate other possible mechanisms, we evaluate the dependences of the RIL and iRIL versus the proton flux and fluence, regardless of their energies. Results are plotted in Figure 9 for the Ce-doped sample but comparable results are observed for the Cu-doped ones, even if they are less impressive as the iRIL and RIL behaviors against equivalent dose and dose rate are less dependent on proton energy. These results unambiguously reveal that the RIL generation mechanism is not directly related to an ionization process but is directly related to the flux and fluence of the protons, regardless of their energies (at least in the 35-63 MeV range).

Cu-doped sample Ce-doped sample linear linear sensitivity sensitivity coefficient Pearson's r coefficient Pearson's r 63MeV

0.0392

0.99706

0.00637

0.99979

57MeV

0.04053

0.99977

0.0055

0.99937

48MeV

0.03157

0.99983

0.00489

0.99939

35MeV

0.0343

0.99984

0.00424

0.99986

0.04

0.02

0.00 35

40

45

50

55

60

65

Proton Energy (MeV) Fig.8. Illustration of the proton energy dependence of the RIL sensitivity coefficients for the Cu- and Ce-doped materials in the 35 –63 MeV range. Tests were performed at RT, varying the beam current from 1 to 7 nA.

For Ce-doped glass, this coefficient seems to slightly increase with the proton energy. A possible explanation could be that the slightly different geometries of our samples (rod sizes of 0.75 and 1 mm for Cu and Ce, respectively) could affect the equivalent dose deposited at the various energies. To investigate the amplitude of the needed correction, calculations have been performed with the Geant4 toolkit version 10.1 [15,16] to check whether the geometry of our samples (thin rod surrounded by a pure silica glass needle) can affect the dose deposition within the doped rod at the various energies of our tests. For this, the actual and slightly different geometries of the two samples have been considered as well as the doping of the rod with either Cu or Ce (about 250ppm of Cu and 0.07 wt% of Ce, deduced from Electron MicroProbe Analysis). The main simulation results are listed in Table I, in which we compare the energy deposition at 35 MeV, 48 MeV and 57 MeV compared to the one at 63 MeV. From these calculations, it appears that the observed influence of proton energy on the RIL dose rate dependence cannot be explained by difference in deposited energy due to our sample geometry.

0.10

RIL (arb. units)

0.08 0.06 0.04 63MeV 48MeV 57MeV 35MeV

0.02

0

2x106

4x106

6x106

8x106

1x107

Proton flux (p/(cm².s-1))

iRIL(arb. units)

Cu-doped sample Edep Comp 1 ref 1.018 ~+2% 1.029 ~+3% 1.047 ~+5%

y= 8.28×10-9x (Pearson's r=0.99861)

0.00

TABLE 1 – COMPARISON BETWEEN THE DEPOSITED ENERGIES BY PROTONS RANGING FROM 35 TO 63 MEV IN CU OR CE-DOPED SILICA Energy 63 MeV 57 MeV 48 MeV 35 MeV

a) Ce-doped sample

Ce-doped sample Edep Comp 1 ref 1.015 ~+1.5% 1.028 ~+2.8% 1.055 ~+5.5%

20 18 16

b) Ce-doped sample

14 12 10

y=8.625×10-9x (Pearson's r=0.99928)

8 6 4 2 0

0.0

Indeed, the equivalent dose deposited are close for all proton energies. The deposited energy is in fact about 5% larger at 35 MeV than at 63 MeV, this difference being the largest one found. This behavior seems opposite to the measured RIL sensitivity decrease with the proton energy (e.g. from 3.92×10-2 to 3.43×10-2 when energy varies from 63 MeV to 35 MeV in Cu-doped sample, namely a 13% decrease). Then, this energy dependence is not related to experimental conditions, but to other phenomena.

35 MeV 48 MeV 57 MeV 63 MeV

5.0x108 1.0x109 1.5x109 2.0x109 2.5x109

Proton fluence (p/cm²) Fig.9. (a) Evolution of the radiation-induced luminescence (RIL) with the proton flux (b) Evolution of the integrated radiation-induced luminescence (iRIL) with the proton fluences. Protons have energies ranging from 35 to 63 MeV and beam currents vary from 1 to 7nA.

From a pragmatic point of view, these two figures confirm that the doped materials elaborated by sol-gel are very efficient systems to count the proton flux and fluence in the 35 to 63 MeV energy range. Even if the exact mechanisms at the origin of the RIL (and OSL) have to be identified, this preliminary work confirms the potential of such sol-gel glasses for proton beam monitoring during protontherapy treatments.

D. Origin of the Proton Energy Influence on RIL Amplitude and Integral: role of ionizing processes The observed influence of the proton energy on the dose and dose rate evaluation through RIL is quite surprising as such



induced attenuation during the irradiation and recovery of plastic optical fibres”. Opt. Laser Techn. , 47, 148-151, 2013. [9] S. Rizzolo, A. Boukenter, T. Allanche, J. Périsse, G. Bouwmans, H. El Hamzaoui, L. Bigot, Y. Ouerdane, M. Cannas, M. Bouazaoui, J-R Macé, S. Bauer and S. Girard, “Optical Frequency Domain Reflectometer Distributed Sensing Using Microstructured Pure Silica Optical Fibers under Radiations”, IEEE Trans. Nucl. Sci., vol.63, n°4, pp. 2038 - 2045, 2016.. [10] H. El Hamzaoui, Y. Ouerdane, L. Bigot, G. Bouwmans, B. Capoen, A. Boukenter, S. Girard, M. Bouazaoui, “Sol-gel derived ionic copperdoped microstructured optical fiber: a potential selective ultraviolet radiation dosimeter”, Opt. Express 20 (28), 29751, 2012. [11] H. El Hamzaoui, G. Bouwmans, B. Capoen, Y. Ouerdane, G. Chadeyron, R. Mahiou, S. Girard, A. Boukenter, M. Bouazaoui, “Effects of densification atmosphere on optical properties of ionic copper-activated sol-gel silica glass : towards an efficient radiation dosimeter”, Mater. Res. Express 1, 026203, 2014. [12] H. El Hamzaoui, B. Capoen, N. Al Helou, G. Bouwmans, Y. Ouerdane, A. Boukenter, S. Girard, C. Marcandella, O. Duhamel, G. Chadeyron, R. Mahiou, M. Bouazaoui, “Cerium-activated sol-gel silica glasses for radiation dosimetry in harsh environment”, Mater. Res. Express 3, 046201, 2016. [13] B. Capoen, H. El Hamzaoui, M. Bouazaoui, Y. Ouerdane, A. Boukenter, S. Girard, C. Marcandella, and O. Duhamel, “Sol–gel derived copper-doped silica glass as a sensitive material for X-ray beam dosimetry” Opt. Mater., vol. 51, 104, 2016. [14] www.triumf.ca/pif-nif [15] S. Agostinelli, et al., "GEANT4 - A simulation toolkit", Nuclear Instruments and Methods in Physics Research A, vol. 506, pp. 250 - 303, 2003. [16] J. Allison, et al., "Geant4 developments and applications", IEEE Transactions on Nuclear Science, vol. 53, pp. 270 - 278, 2006. [17] S. Rizzolo, A. Boukenter, E. Marin, T. Robin, M. Cannas, A. Morana, J. Périsse, J-R Macé, Y. Ouerdane, B. Nacir, P. Paillet, C. Marcandella, M. Gaillardin, and S. Girard, “Evaluation of Distributed OFDR-based Sensing Performances in Mixed Neutron/Gamma Radiation Environments”, presented at NSREC 2016, submitted to IEEE Trans. Nucl. Sci., 2016.

V. CONCLUSIONS We present very innovative silica sol-gel glasses doped with emitting Cu+ or Ce3+ ions embedded in the radiation resistant a-SiO2 matrix and their optical responses under proton exposures. The two types of glasses exhibit both interesting OSL and RIL signatures, opening the way to the development of new online dosimeters for protontherapy applications. The monitoring of the RIL during the exposure allows to follow, with a resolution time of a few ms, the evolution of the proton flux regardless of their energies (in the investigated range from 35 to 63 MeV). Knowing the conversion factor in the target material, it is possible to obtain the equivalent dose rates at the various energies. By integrating the RIL signal over the whole exposure, it is possible to precisely estimate the proton fluence (and then the equivalent deposited dose). Results are more complex for OSL measurements and, even if it seems possible to exploit this signature for dose reconstruction, more experiments are needed to evaluate the potential of this approach. From a basic mechanisms point of view, it is very interesting to notice that our results demonstrate that the RIL phenomena is not strictly related to ionization processes, as it could be expected from previous studies on silica, but is rather directly related to the proton flux and fluence. This will be deeply investigated in the near future to identify the mechanism at stake under proton exposure and compare them to the ones observed under X- or gammarays. ACKNOWLEDGMENT The authors thank TRIUMF’s committee for providing the proton beam time. This work was partially supported by the Nord-Pas de Calais Regional Council through the Contrat de Projets Etat Region (CPER) “Photonics for Society”, the FLUX Equipex Project (ANR-11-EQPX-0017) and the Labex CEMPI (ANR-11-LABX-0007). REFERENCES [1] S. Girard, J. Kuhnhenn, A. Gusarov, B. Brichard, M. Van Uffelen, Y. Ouerdane, A. Boukenter, and C. Marcandella, “Radiation Effects on Silicabased Optical Fibers: Recent Advances and Future Challenges”, IEEE Trans. Nucl. Sci., vol.60, n°3, pp. 2015 - 2036, 2013. [2] D. Di Francesca, S. Girard, S. Agnello, C. Marcandella, P. Paillet, A. Boukenter, F.M. Gelardi, and Y. Ouerdane, “Near infrared radioluminescence of O2 loaded Rad-Hard silica optical fibers: a candidate dosimeter for harsh environments”, Appl. Phys. Lett., vol.105, 183508, 2014. [3] KW Jang, WJ Yoo, SH Shin, D Shin, B. Lee “Fiber-optic Cerenkov radiation sensor for proton therapy dosimetry”, Opt. Express., vol. 1820(13), 13907-14, 2012. [4] S. O’Keeffe et al., “A review of recent advances in optical fibre sensors for in vivo dosimetry during radiotherapy” Brit. J. Radiol.,vol. 88, no. 1050,art. no. 20140702, 2015. [5] M. Benabdesselam, F. Mady, S. Girard, Y. Mebrouk, J.B Duchez, M. Gaillardin, P. Paillet, “Performance of Ge-doped Optical Fiber as a Thermoluminescent”, IEEE Trans. Nucl. Sci., vol.60, n°6, pp. 4251-4256, 2013. [6] S. O’Keeffe, W. Zhao, W. Sun, D. Zhang, Z. Qin, Z. Chen,Y. Ma, and E. Lewis, “An Optical Fibre-Based Sensor for Real-Time Monitoring of Clinical Linear Accelerator Radiotherapy Delivery”, IEEE J. Sel. Topics Quant. Elect., vol 22, n° 3,5600108 , 2016. [7] E. Janata, M. Körfer, “Radiation Detection by Cerenkov Emission in Optical Fibers at TTF”, Tesla-Report 2000-27, December 2000. [8] M.S. Kovacevic, S. Savovic, A. Djordjevich, J. Bajic, D. Stupar, M., Kovacevic, S. Simic, “Measurements of growth and decay of radiation
