ARTICLE IN PRESS

0168-9002/$ - se

doi:10.1016/j.ni

�CorrespondE-mail addr

Nuclear Instruments and Methods in Physics Research A 572 (2007) 785–793

www.elsevier.com/locate/nima

Assessment of the radiation tolerance of LaBr3:Ce scintillators to solarproton events

Alan Owensa,�, A.J.J. Bosb, S. Brandenburgc, E.-J. Buisd, C. Dathye, P. Dorenbosb,C.W.E. van Eijkb, S. Kraftd, R.W. Ostendorfc, V. Ouspenskie, F. Quaratia

aScience Payload and Advanced Concepts Office, ESA/ESTEC, Noordwijk, The NetherlandsbTechnical University of Delft, Mekelweg 15, Delft, The Netherlands

cKernfysisch Versneller Instituut, University of Groningen, Zernikelaan 25, 9747 AA Groningen, The NetherlandsdCosine Research BV, Niels Bohrweg 11, 2333 CA Leiden, The Netherlands

eSaint-Gobain Cristaux 104 Route de Larchant BP 521, 77794 Nemours CEDEX, France

Received 22 September 2006; received in revised form 6 December 2006; accepted 7 December 2006

Available online 17 January 2007

Abstract

Radiation effects caused by solar proton events will be a common problem for many types of sensors on missions to the inner solar

system because of the long cruise phases coupled with the inverse square scaling of solar particle events. In support of the BepiColombo

and Solar Orbiter missions we have undertaken a comprehensive series of tests to assess the effects on a wide range of sensors. Sensors

based on scintillators have been proposed for both missions. In this paper, we report on a series of controlled irradiations on a set of five

LaBr3:Ce scintillators. The crystals are 1 in. right circular cylinders of LaBr3, packaged in aluminium housings and viewed through BK7

optical windows. Four crystals were exposed to simulated solar proton events over the energy range 60–200MeV having a spectral shape

approximating that of the August 1972 solar particle event. Each crystal was exposed to a different total fluence. One crystal was exposed

to an integral fluence of 109 protons cm�2, a second to 1010 protons cm�2, the third to 1011 protons cm�2 and the fourth to

1012 protons cm�2. The latter corresponds to an absorbed dose in silicon of 1Mrad or in SI units, 10 kGy. The fifth crystal served as a

reference. The crystals were characterized both before and after the irradiations in terms of energy resolution, light output and

background count rate. The key conclusions of the study are that LaBr3 is radiation tolerant showing no measurable degradation effects

when exposed to experimentally simulated solar proton flare spectra with fluences up to 1012 protons cm�2 (�1Mrad or 10 kGy

equivalent in silicon) and integrated above a 60MeV energy threshold. LaBr3 behaves as a generic intermediate mass material showing

similar activation yields as CsI(Tl) and Ge above 1010 protons cm�2 (�10 krad (100Gy) equivalent in silicon) and significantly less than

Ge below this fluence.

r 2007 Elsevier B.V. All rights reserved.

PACS: 07.85; 07:89.+b; 29:40.Mc

Keywords: Lanthanum–bromide; Scintillation detectors radiation damage; g-rays

1. Introduction

The recently discovered cerium-doped lanthanum halide(LaX3:Ce) scintillators could potentially revolutionizespectroscopic systems because of their dramatically im-proved light yield, linearity and speed compared to

e front matter r 2007 Elsevier B.V. All rights reserved.

ma.2006.12.008

ing author. Tel.: +3171 565 5326; fax: +31 71 565 4690.

ess: aowens@rssd.esa.int (A. Owens).

traditional scintillators. However, their development is sorecent that many of their physical properties and particu-larly, their radiation tolerance and propensity to activateare unknown.For inner planetary missions, proton radiation damage

is potentially a very serious problem in view of long cruisephases and the inverse square scaling of solar particle fluxeswith distance to the Sun. Radiation damage studies to datehave concentrated on emulating the galactic cosmic rays

ARTICLE IN PRESS

Fig. 1. A composite of solar proton spectra (adapted from Ref. [2]). The

present work considered the August 1972 event. Note, we have not shown

the anomalous Bastille day event since it is an order of magnitude larger

and had a very hard spectrum during the peak emission.

A. Owens et al. / Nuclear Instruments and Methods in Physics Research A 572 (2007) 785–793786

using GeV protons with fluences in the range 107–109 pro-tons cm�2, since these are the most relevant to missions at1AU or greater. These fluences correspond to absorbeddoses in silicon of �10 rads–1 krad, or in SI units0.1–10Gy.1 For these conditions, it has been successfullydemonstrated that scintillating materials show little evi-dence of damage effects. However, what was not clear iswhether scintillator detectors would suffer significantdamage in the inner solar system purely due to the inversesquare scaling. Here, solar particle events may be expectedto reach fluences of up 1010 protons cm�2 or more for asingle event, which should be compared to the averageannual cosmic ray fluence of �108 protons cm�2. Inaddition to the obvious question of degradation of thespectral resolution, secondary issues such as activation alsoneed to be addressed and assessed.

2. Radiation environment

For inner heliospheric missions one has to consider twocomponents of radiation—the galactic cosmic rays (GCRs)and solar energetic particles (SEP). Whereas the GCRs dueto their large interaction depth induce most local produc-tion throughout the spacecraft, it is the smallest in terms oftotal dose. Particle energies tend to be in the range0.1–10GeV and integral particle fluxes range from 2 to4 cm�2 s�1, depending on the position in the solarcycle. This corresponds to an average annual fluenceof 2� 108 cm�2. SEP events on the other hand, present amuch more serious threat. During periods of solaractivity the Sun can produce copious particle emission,mainly protons. Events can last for days and fluxes canvary by as much as six orders of magnitude [1], potentiallyreaching fluences of 1011 cm2 or more near the orbit ofMercury. Peak fluxes can stay high for periods of hours.Particle spectra tend to be flat, rolling over at about30MeV, but providing significant fluxes up to a fewhundred MeV.

Radiation damage assessments usually proceed byassuming an average over the solar cycle or including onemajor solar flare of the August 1972 magnitude. Since solarevents are stochastic in nature, the probability of exceedinga particular fluence can only be realistically assessed on astatistical basis. For example, at the Earth, there is roughlya 15% chance per annum of encountering at least one solarflare of total fluence 4109 protons cm�2 and a 6% chanceof encountering at least one with a fluence 41010 proto-ns cm�2, for fluences integrated from 30MeV upwards.However, at the orbit of, say, Mercury fluences will benearly an order of magnitude more intense if an inversesquare scaling is assumed. Thus, a 109 protons cm�2 eventat the Earth is a 1010 protons cm�2 event at Mercury andwhen assessing the radiation tolerance of new detectormaterials for inner solar system applications, one shouldfollow the same design practices adopted for Earth orbit,

11Gy ¼ 100 rads.

scaled to the orbit of Mercury. In this case, a flare size of1010 protons cm�2 is considered a canonical large flare atthe Earth and therefore tests should proceed to a maximumfluence of at least 1011 protons cm�2.We report on a series of controlled irradiations on four

packaged LaBr3 crystals with a volume of the order of10 cm3 using a simulated solar proton spectrum ofincreasing fluence, up to a maximum fluence of 1012 pro-tons cm�2 and a spectrum approximating the shape of the1972 flare. The choice of this particular flare can beunderstood with reference to Fig. 1, in which we plot thespectral distributions of a number of large solar protonevents. From the figure, we see that the August 1972 flare isa good average, in that it has balanced distributions at bothbelow and above 100MeV.

3. Radiation damage in scintillators

In g-ray detectors, radiation damage manifests itself asan alteration of the operational and detection propertiesand is mainly caused by the interactions of high-energyparticles with the bulk material. The dominant source ofdamage is proton induced, largely due to their large fluxand interaction cross-section. Neutrons resulting fromintranuclear cascades in the detector or spacecraft systemsalso contribute significantly, but usually at a level ofapproximately 10% of that of the protons.In inorganic scintillators, the creation of radiation

damage is a complicated process, which involves not onlythe host crystal but also its impurities and defects. Ingeneral, damage manifests itself in four ways:

(1)

The formation of color centre bands that absorbphotons emitted by the luminescence centres, causinga decrease in optical transmission. These reduce the

ARTICLE IN PRESS

2S

A. Owens et al. / Nuclear Instruments and Methods in Physics Research A 572 (2007) 785–793 787

attenuation length of scintillation photons in the crystalthereby reducing the light output and ultimately theresolution. The obvious physical manifestation of thisprocess is a cloudiness or discoloration of the crystal.

(2)

Direct damage to the luminescent centres causing achange in the emission characteristics (e.g., efficiency,decay time). The luminescent centres may be renderedinoperable (quenching) due to valance changes orcompensatory ion diffusion. Degradation in lightoutput can then occur because of a reduction in theenergy transfer from the electrons and/or holes to theluminescent centres following the creation of pointdefects and traps. The traps also reduce carriermobilities.

(3)

The creation of shallow traps, which increase theafterglow (phosphorescence) of the crystal addingadditional noise to the scintillation signal.

(4)

The creation of radionuclides produce a form of noise,which can completely swamp the signal in highlyactivated scintillators.

Table 1

Summary of pre-irradiation resolution measurement results

Detector

no.

Detector

bias volts

Energy (keV)

60 662 1173 1332

FWHM energy resolution (%)

J149 680 10.6 2.8 2.0 2.0

J150 680 12.0 3.3 2.6 2.4

J146 700 11.0 3.1 2.4 2.2

In general, noticeable effects are only apparent forabsorbed doses in the 10 kGy (Mrad) range and forthis reason, scintillators are generally considered radiationhard, when compared to other sensor materials (e.g.,semiconductors). Even if damage occurs, thermal anneal-ing and optical bleaching can be effective in restoringperformance. However, as noted above, one potentialproblem for scintillators is the activation of its constituentelements following high-energy particle interactions.While all materials activate to some degree, the increasinguse of very high Z materials in scintillating compounds,such as lutetium and lanthanum, meant to ensure a high-interaction probability, also increase the propensity toactivate. In addition high Z elements tend to have naturallyoccurring radioactive isotopes, which add a steady-statebackground to the scintillation signal. At the present time,there is little or no experimental data available onlanthanum halides.

4. Experimental

The tests were carried out on a set of five packagedcerium-doped LaBr3 scintillation crystals produced bySaint Gobain Crystals and Detectors. The Ce fractionwas 5%. The crystals are right circular cylinders ofdiameter 25.4mm and length 25.4mm. To detect as muchscintillation light as possible the crystals are covered on allsides, except the read-out side, with a proprietary reflectivematerial.2 This takes the form of a foil, which is held inplace by tape which itself is reflecting. The scintillation lightis extracted through a 30mm diameter, 3mm thick BK7‘‘grade A’’ glass window providing good optical perfor-mance over the UV and near-infrared wavelength region(flatness: l/2 @ 633 nm, wavefront distortion: l 25mm�1).

aint Gobain Crystals and Detectors.

The entire assembly is then hermetically sealed in analuminium container to prevent hydration. The aluminiumcan is 0.5mm thick, which imposes a �20 keV low-energythreshold.The scintillation signal is detected and amplified by a

Photonis XP2060B, 39mm diameter, photo multiplier tube(PMT). The PMTs entrance window is optically coupled tothe scintillator by optical grease. The XP2060 has 10dynode stages and a bialkali photocathode, which ensuresa wide wavelength response and good pulse heightresolution and linearity at low biases. Low-voltage opera-tion is necessary because of the very high scintillation lightoutput of LaBr3 (60,000 photonsMeV�1) coupled with itsshort duration (t�15 ns).The scintillators were irradiated without PMTs in

order to decouple radiation effects in the crystal assemblyfrom the readout sensor. Therefore, the PMTs werepackaged into separate assemblies that could beeasily attached to the crystal assemblies. To complete theentire detector assembly, a PMT base assembly, whichincorporates an integral low-noise preamplifier and voltagedivider network, plugs directly into the PMT assembly.The entire ensemble is modular consisting of interchange-able parts allowing measurements to be made on-the-fly atradiation facilities. The rest of the analog chain consists ofan Ortec 671 spectroscopy amplifier whose output isdigitized by an Amptek MCA8000A 12-bit ADC andstored on a PC. The shaping time used throughout themeasurements was 3 ms. Detector bias is provided by aCanberra 3106D high-voltage supply unit. The biasvoltages used by the detector assemblies were typicallyaround +700V, which gave optimum spectral resolutionsof �3% FWHM at 662 keV.

4.1. Pre-irradiation laboratory measurements

The crystal performances were assessed using a series ofradioactive sources (241Am, 137Cs and 60Co) placedcoaxially above the forward detection face. The pre-irradiation energy resolutions of all five crystals aresummarized in Table 1. The resolutions are typicallyaround 3.0% FWHM at 662 keV and 2.2% FWHM at1332 keV.

J148 700 10.6 3.0 2.2 2.1

J147 700 11.0 3.0 2.4 2.2

ARTICLE IN PRESS

Fig. 2. Pre-irradiation 137Cs and 60Co spectra acquired with detector J148. All scintillation crystals gave similar responses. The bias voltage was 700V and

the amplifier shaping time 3ms. The FWHM energy resolutions are 3% at 662keV and 2.1% at 1332 keV, respectively.

Fig. 3. The background radiation produced in and detected by a 12.8 cm3 LaBr3 crystal. (a) The background measured externally by a 120 cm3 high-purity

Ge detector in a Pb castle and (b) laboratory background measured directly in the LaBr3 crystal by a PMT.

Table 2

The steady-state volume activity yields for the two principal activities

emanating from LaBr3

Isotope Energy (keV) Volume activity yield

(photons cm�3 s�1)

138La 789 0.6425138La 1436 0.9289

A. Owens et al. / Nuclear Instruments and Methods in Physics Research A 572 (2007) 785–793788

Representative 137Cs and 60Co spectra are shown inFig. 2 and are classical g-ray energy-loss spectra with well-resolved full-energy peaks, Compton edges and backscatterfeatures. The uniformity of both the spatial and angularresponses was then assessed using a collimated 137Cs sourceplaced at various positions around the crystal. Novariation was found within statistics.

The background intrinsic to LaBr3 and the extrinsicnatural radiation background measured by LaBr3,have been investigated both externally using a large-volume high-purity Ge detector and internally by readingout the scintillator with a PMT. The purpose ofthe measurements was to evaluate the level of spectralcontamination by radioactive isotopes of lanthanum and inparticular 138La, which constitutes 0.09% of naturallyoccurring lanthanum (139La). The external measurementswere carried out using an ultra-low background Pbcastle to suppress the normal laboratory background.A LaBr3 crystal assembly was placed on top of theGe detector and a spectrum accumulated. The resultantspectrum is shown in Fig. 3 (left) from which we seethat LaBr3 emits primarily two lines at 789 and 1436 keV.These have been attributed to 138La. Weaker lines at 236,256, 271, 351, 405, 427 and 1766 keV are attributed to thedecay of daughter products of the actinium series (Ac is a

known contaminant of lanthanum halide crystals). Inaddition, the ubiquitous annihilation line at 511 keV, the40K line at 1460 keV and the 208Tl line at 2615 keV are alsopresent. The associated continuum is consistent with theexpected Compton distribution from these lines. Using theGEANT 4 electron–photon hadron code we have deter-mined the energy-dependent solid angle factor andconverted the line fluxes derived from Fig. 3(a) intosteady-state volume production yields in LaBr3. These arelisted in Table 2. The total count rate of1.57 count cm�3 s�1 is in agreement with that expected(1.51 count cm�3 s�1), assuming the abundance of 138La innatural La to be 0.09%. Additionally, the measured countrate in the two lines matches the branching ratio of the twodecay modes.

ARTICLE IN PRESSA. Owens et al. / Nuclear Instruments and Methods in Physics Research A 572 (2007) 785–793 789

Lastly, in Fig. 3(b) we show the laboratory back-ground recorded over a 60 h period in the referenceLaBr3 crystal (J149). From the figure, we see anumber of line features, most of which can be attributedto intrinsic activation due to a-decay. The total countrate of this natural radioactivity including back-ground from the laboratory was �4 count cm�3 s�1 oralternately 4Bq cm�3.

5. Proton irradiations

Proton irradiations were carried out at the KernfysischVersneller Instituut in Groningen, The Netherlands. A190MeV proton beam was extracted from the AGORsuper-conducting cyclotron [3] and transported to theirradiation hall whereupon it was expanded and spatiallylinearized using a dual scatterfoil method [4]. The firstflat scatterfoil produces a Gaussian radial flux distributionand the second inhomogeneous foil produces an overallflat dose distribution. The total scatterer thickness isabout 2mm, which corresponds to an energy loss ofthe incident protons of only 5MeV. Brass collimatorsare then used to define the boundaries of the irradiationfield, which is up to 8 cm diameter at the target.The measured non-uniformity across this field is o2%.A schematic of the beamline layout is shown in Fig. 4.The LaBr3 crystals were positioned in this field using athree-axis laser-surveying system, capable of locating anobject in space with an accuracy of a fraction of amillimeter. The crystals were positioned in such a way thattheir windows directly faced the beam to ensure that theBragg peaks at low and intermediate energies peak are notlocated in the glass. Note, that for the size of the crystalrequired for the BepiColombo and Solar Orbiter spectro-meters, these energies (and their much increased fluxesbecause of the exponential shape of the incident protonspectrum) would normally terminate in the scintillator andnot the glass.

The incident particle fluence on the detector wasdetermined using a combination of three techniques from

Fig. 4. Schematic representation of the radiobiology beamline layout at

the Kernfysisch Versneller Instituut (see text).

which two were absolute methods using scintillationdetectors and activation measurements of aluminiumfoils and another a relative method using a calibratedion chamber. The activation measurements rely onthe 27Al(p,x)24Na reaction which produces a de-excitationline at 1368 keV. Since the cross-sections for this re-action have been accurately measured across the energyrange 60–200MeV [5], the line intensity gives a directmeasure of the incident proton flux. The neutron-inducedproduction of 24Na results in a 3% increase of the intensity,which has been corrected for. Combining the results fromthe three methods, the absolute accuracy in the fluencedetermination is 10%. The total neutron production at thebeamline is estimated to be around 30% of the number ofprotons sent into the beamhall. Around 10% ofthese are produced in the collimator defining the protonfield, the remainder in various collimators upstream(distances between 1 and 3m). The maximum secondaryneutron flux at the detector was estimated to be �10% ofthe proton flux.

5.1. Proton energy spectrum

The incident spectrum at the detector was shaped usingan energy degrader consisting of a set of 1 and 4mm thickcopper plates. Fine tuning at high energies was achievedusing polystyrene plates. Copper was chosen over plastic orwater for the primary degrading medium, because itensures a more uniform irradiation field in the vicinity ofthe target. Combinations of plates were introducedadditively to replicate the spectrum at seven energies;namely 61, 80, 100, 120, 140, 160 and 184MeV. The centreenergy in each band could be determined to a precision ofp2MeV.The energy range was chosen as follows. The lowest

energy was chosen to be representative of the energythreshold imposed by a few g cm�2 of shielding afforded bythe spacecraft. The upper threshold was chosen because atthis energy, SEP spectral fluxes generally fall to �1% oftheir values at 10MeV (see Fig. 1). The energy spread inthe beam has been determined from Monte Carlo calcula-tions and is reasonably Gaussian. The energy spread of thebeam is determined by straggling in the degrader, andranges from 15% FWHM at 61MeV too1% FWHM at184MeV.

5.2. The irradiations

The irradiations took between 60 s and 1 h depending onthe required fluence. The irradiation rate ranged between106 and a few times 108 protons cm�2 s�1. Representativesub-components of the scintillator assembly (the housing,reflector and window) were also irradiated separately inorder to estimate their contribution to the overall energyresolution degradation and background in the spaceenvironment.

ARTICLE IN PRESSA. Owens et al. / Nuclear Instruments and Methods in Physics Research A 572 (2007) 785–793790

6. Results

6.1. Spectra

The results of the irradiations are given in Table 3. Thesedata were taken once the crystals were back in ourlaboratory using a custom readout assembly with a51mm diameter ETL 9266 PMT. This allowed possibledamage-related shifts in the source peak positions to beeasily determined since the bias and amplifier conditionscould be kept constant for all samples. However, while thedynamic range of this tube was better than that achievedwith the XP 2060, the linearity was worse and this isreflected in the slightly broader energy resolutions for theunirradiated sample. Compilations of 137Cs and 60Cospectra for each irradiation are shown in Fig. 5, fromwhich we see that apart for the appearance of activationlines at the higher fluences, no obvious change in spectra isapparent for fluences below 1011 protons cm�2. In fact forthe energy ranges shown, only �4 activation lines featurestrongly at the higher fluences, at energies of 80, 130, 329and 559 keV. Note, all the spectra shown in Fig. 5 weretaken at the same time after the irradiation and theordinates on the graphs have the same range for the firstfour doses. For the 1012 protons cm�2 graphs, the y-axishas been increased by 25% to allow for the apparent rise inthe peak height due to the increased continuum caused byactivation. The samples were examined physically aftereach irradiation. At a fluence level of 1011 protons cm�2 avery slight discoloration of the scintillator could bediscerned, although there was no statistically significantchange in energy resolution. In comparison with pre-irradiation data an apparent 20% change in gain hadoccurred, or more specifically the centroid of the 662 and1332 keV calibration peaks had moved to lower channelsby �20%. More strikingly, the sample exposed to1012 protons cm�2 was clearly discolored, the energyresolution had degraded by �40% and the gain reducedby �50%. However, it was also noted that the window

Table 3

Summary of post-irradiation resolution measurements

Detector

no.

Fluence

Protons cm�2Energy (keV)

60 662 1173 1332

FWHM energy resolution %

J149 0 10.6 3.4 (2.8) 2.6 (2.0) 2.5 (2.0)

J150 109 12.0 3.4 (3.3) 2.9 (2.6) 2.6 (2.4)

J146 1010 10.6 3.9 (3.1) 2.8 (2.4) 2.6 (2.2)

J148 1011 11.6 3.5 (3.0) 2.7 (2.2) 2.7 (2.1)

J147 1012 12.1 4.0 (3.1) 3.5 (2.4) 3.5 (2.2)

J147a 1012 12.2*(11.0) 3.2*(3.1) 2.3*(2.4) 2.4a (2.2)

The pre-irradiation values are shown in parentheses. The absolute error on

these measurements is estimated to be 70.2%.aWindow replaced and resolutions re-measured.

sample irradiated to the same fluence had discolored by thesame amount. A test in which the un-irradiated referencecrystal was viewed through the un-irradiated referencewindow and then through the 1012 protons cm�2 irradiatedwindow reproduced the resolution measured in theirradiated sample as well as the gain shift. The crystalwas subsequently repackaged with a new window and theenergy resolution and gain were found to be the same asthe pre-irradiation values within statistics. Optical mea-surements were also carried out on the subcomponents,namely the transmission of the window and reflectance’s ofthe reflector foil and tape. Additional irradiations atfluences of 1011 and 1012 protons cm�2 were later carriedout on two quartz windows. We summarize the results asfollows:

(1)

there is no measurable degradation in the scintillationproperties of the irradiated crystals,

(2)

there was no measurable change in the reflectance ofthe reflector system,

(3)

there was a loss of light output in the BK7 opticalwindows. This amounted to a transmission loss at380 nm of �40% for an irradiating fluence of 1012 pro-tons cm�2 and

(4)

no transmission loss was measured in quartz windowsfor fluences up to 1012 protons cm�2.

6.2. Activation

After the irradiations, the count rates due to activationwere monitored as a function of time. Because of the largeactivation of the crystal exposed to 1012 protons cm�2, itcould not be handled for several weeks. Initially, after theirradiation the registered activity was 2mSv/h. After 15days the activity fell to 0.2mSv/h. In Fig. 6 we show thedecay of the spectra accumulated in the 0.2–8MeV bandsome 18, 46, 78 and 450 h after exposure. We see thatessentially all of the activation occurs at energies less than3MeV. However, it should be noted that the ordinatecovers eight orders of magnitude and so the apparentenergy range over which activation dominates is mislead-ing. In Fig. 6(b) we show the integral distributions of Fig.6(a)—specifically, the fraction of counts that occur belowan energy E. We see that �1 day after a hypothetical1012 protons cm�2 flare, 90% of events due to activationoccur at energies less than 1100 keV, whereas 16 days laterthe same fraction of events occur at energies o700 keVindicating a softening of the spectrum. In both cases, thereare virtually no events above 2MeV.In Fig. 7 we show the decay curves for the 1010 and

1012 protons cm�2 irradiated samples using the same dataused to produce Fig. 6. The count rates represent theintegral counts over the energy range 200 keV–8MeV.Using power law extrapolations for both data sets, weestimate that the background will return to within a factorof two of its original pre-irradiation value within �12 and50 days, when irradiated at 1010 and 1012 protons cm�2,

ARTICLE IN PRESS

Fig. 5. Compilation of 137Cs and 60Co spectra taken 5 days after each irradiation. The irradiating fluences in units of protons cm�2 are given between

spectra. Note the y-axis is the same for the first four spectra and the only obvious effect of the irradiations is the growth of activation lines at the highest

fluences.

A. Owens et al. / Nuclear Instruments and Methods in Physics Research A 572 (2007) 785–793 791

respectively. Bear in mind however, as can be seen fromFig. 6, 90% of this increased background will occur atenergies o700 keV.

In Fig. 8, we show the measured count rates in theenergy region 0–3MeV as a function of irradiated fluence,

measured by all four crystals 17 days after the irradiations(450 h). The count rates are given in the form of activationvolume yields (i.e., the number of counts in the 0–3MeVenergy range that can be attributed to the activation of thesamples in units of Bq cm�3—in essence we take the count

ARTICLE IN PRESS

Fig. 6. (a) Background spectra measured 18, 46, 78 and 450h after an exposure to a proton fluence of 1012 protons cm�2and (b) the fraction of the total

counts measured below an energy E for background spectra measured 18, 48, 78 and 384h after irradiation. The intersection of the curves and the dotted

line allows us to determine the energy below which 90% of the activation is contained.

Fig. 7. Decay profiles for the 1010 and 1012 protons cm�2 data used to

produce Fig. 6. The count rates are integrated over the energy range

0.2–8MeV. The dashed lines show power-law extrapolations to the data.

For comparison, we also show the pre-irradiation background level.

Fig. 8. The volume yields due to activation (i.e., the number of counts in

the 0–3MeV energy range that can be attributed to the activation of the

samples in Bq cm�3). The yields are given for different irradiation fluences

and materials. Note the intrinsic activities of the crystals have been

subtracted—so the measured yields are strictly due to the activation of the

sample due to the incident fluence. The data were all taken 450h after

irradiation. Note, the Ge data point at 6� 1010 protons cm�2 is a lower

limit since the efficiency of the detector had been seriously reduced by

radiation damage.

A. Owens et al. / Nuclear Instruments and Methods in Physics Research A 572 (2007) 785–793792

rate at time t after the irradiation, subtract the pre-irradiation background and divide by the crystal volume).For comparison, at 1010 protons cm�2 we also plot thederived yields for a CsI(Tl) crystal and a germanium crystalexposed to the same fluence. In fact LaBr3 behaves as ageneric intermediate mass material showing the sameactivation yields as CsI(Tl) and Ge above 1010 proto-ns cm�2 and significantly less than Ge below this fluence.Note, the Ge data point at 6� 1010 protons cm�2 is a lowerlimit because the damage induced by this fluence was foundto have radically reduced the efficiency of the detector. Fora given incident proton fluence, F, the activation yields, Ca,of LaBr3 can be well described by a simple empiricalformula

Ca ¼ 1:1� 10�10F ðprotons cm�2Þ (1)

in units of Bq cm�3.A comprehensive activation analysis of the g-ray

emissions from one of the crystal samples (J148) was

carried out using a lead castle equipped with a large volume(198 cm3) Ge detector. The sample had been irradiated witha fluence of 1011 protons cm�2. The spectrum measured50 h after the irradiation revealed the presence of �120discrete lines across the energy range 30 keV–3 MeV ofwhich about a half have been identified and of these about40% can be attributed to La or Br. The results of thisanalysis are presented elsewhere [6].Lastly, we investigated how activation lines manifest

themselves when operating a LaBr3 crystal as a spectro-meter. In Fig. 9 we show a compilation of irradiatedspectra acquired 17 days after the irradiations. Fromthe figure, we see a number of activation lines along withtheir identifications. What is apparent is that significant

ARTICLE IN PRESS

Fig. 9. Compilation of irradiated spectra acquired �17 days after the

irradiations, along with a tentative identification of activation line

features. For comparison we also show the un-irradiated spectrum. Note,

we have not subtracted the intrinsic background of the crystals—hence the

pre-irradiated, 109 and 1010 protons cm�2 spectra are similar since the

intrinsic background is much greater than the induced count rate due to

activation in these samples.

A. Owens et al. / Nuclear Instruments and Methods in Physics Research A 572 (2007) 785–793 793

activation features only appear for irradiating fluencesX1011 cm�2.

7. Discussion and conclusions

We conclude from this study that LaBr3 is radiationtolerant showing no measurable degradation effects

when exposed to simulated solar proton flare spectrawith fluences up to 1012 protons cm�2 (�10 kGy or 1Mradabsorbed dose in Si), integrated above a 60MeV thres-hold. In addition, it behaves as a generic intermediatemass material showing the same activation yields inBq cm�3 as CsI(Tl) and Ge above 1010 protons cm�2

(100Gy or 10 krad) and significantly less than Ge belowthis fluence.

Acknowledgement

The SCI-A team acknowledges the support of theMGNS team during the conduct of this extensive radiationtest programme.

References

[1] D. Smart, M. Shea, Adv. Space Res. 30 (2002) 1033.

[2] J. Wilson, J. Miller, in: F.A. Cucinotta (Ed.), Shielding Strategies for

Human Space Exploration, vol. 3360, NASA Conference Publication,

1997, p. 109.

[3] S. Brandenburg, T. Nijboer, H. Post, L.P. Roobol, H.W. Schreuder, S.

van der Veen, in: F. Marti (Ed.), Proceedings of the XVIth

International Conference on Cyclotrons and their Applications,

East-Lansing (USA), AIP Conference Proceedings vol. 600, 2001,

p. 99.

[4] E. Grusell, A. Montelius, A. Brahme, G. Rikner, K. Russell, Phys.

Med. Biol. 39 (1994) 2201.

[5] R. Michel, F. Pfeiffer, R. Stuck, Nucl. Phys. A 441 (1985) 617.

[6] E. -J. Buis, et al., to Nucl. Instr. and Meth., submitted for publication.