Gamma scintillator system using boron carbide for neutron detection

Y. Ben-Galim a, U. Wengrowicz b, A. Raveh c, I. Orion a

a Department of Nuclear Engineering, Ben Gurion University of the Negev, Beer-Sheva, Israelb NRC-Negev, PO Box 9001, Beer-Sheva 84190, Israelc Advanced Coatings Center at Rotem Industries Ltd., Mishor Yamin, D.N. Arava 86800, Israel

a r t i c l e i n f o

Article history:Received 14 January 2014Accepted 11 April 2014Available online 21 April 2014

Keywords:Neutron detectionBoron carbideScintillatorMonte CarloMATLABMCNP

a b s t r a c t

A new approach for neutron detection enhancement to scintillator gamma-ray detectors is suggested. Byusing a scintillator coupled with a boron carbide (B4C) disc, the 478 keV gamma-photon emitted fromthe excited Li in 94% of the 10B(n,α)7Li interactions was detected. This suggests that the performance ofexisting gamma detection systems in Homeland security applications can be improved. In this study,a B4C disc (2 in. diameter, 0.125 in. thick) with �19.8% 10B was used and coupled with a scintillatorgamma-ray detector. In addition, the neutron thermalization moderator was studied in order to be ableto increase the neutron sensitivity. An improvement in the detector which is easy to assemble, affordableand efficient was demonstrated. Furthermore, a tailored Monte-Carlo code written in MATLAB wasdeveloped for validation of the proposed application through efficiency estimation for thermal neutrons.Validation of the code was accomplished by showing that the MATLAB code results were well correlatedto a Monte-Carlo MCNP code results. The measured efficiency of the assembled experimental model wasobserved to be in agreement with both models calculations.

& 2014 Published by Elsevier B.V.

1. Introduction

Neutron detection has an important role in many industriesapplications such as Radiation Portal Monitors (RPMs) for Home-land Security (HLS) [1], energy production in nuclear power plants,cancer therapy [2], and more. Due to the high thermal neutronabsorption cross-section, gas proportional counters based on 3He(5330b), has been widely used as an efficient neutron detector.However, 3He is very rare (0.000137% of natural He) and the mainsource of its production is tritium decay. Still, the production of3He was diminished by the end of the Cold-War and the commer-cial production of 3He has been practically ended. Therefore, it isnecessary to find other alternatives to 3He. 10B is among thepromising candidates due to the high absorption cross-section(3840b). From the neutron–boron reaction 10B(n,α)7Li, He and Li(Q1¼2.31 MeV in 94% of the interactions) ions are emitted. Themean free path of these energetic ions is very short and special gasproportional counters, boron loaded scintillators and boron coatedsemiconductor detectors should be employed in order to measurethe neutron reaction products.

In this study, two models of 2�2 cm2 boron based neutrondetectors were utilized using Monte-Carlo code written in MATLAB:(A) B4C coated semiconducting planar detector for detecting the α

and Li ions; and (B) Coupled scintillating detector with B4C neutronconverting disc for detecting the gamma photon. This is because thecapturing of the 478 keV gamma-photon emitted from the excited Liin 94% offers alternative detection method. For both models, optimalthickness of the B4C coating was estimated for natural occurring 10Bcompound of 19.8% atomic abundance and for 10B enriched to 96%.The coatings thickness was optimized by the Intrinsic ThermalNeutron Efficiency (ITNE) value. The ITNE is defined as the ratiobetween the number of thermal neutrons hitting the absorbingmedia (B4C) to the number of neutron reaction products that reachesto the surface of the secondary particle detector (α and Li iondetector or gamma photon detector).

Regarding model (A), previous studies, (e.g. Refs. [3,4], showedITNE values of �4–5% for planar coated semiconductors. Otheralternatives for neutron detectors based on semiconductors areperforated semiconductor detectors which shows higher efficiencyvalues such as ITNE of 25% for ribbed trench design [5], theoreticalITNE higher than 70% for the 3D matrix of p-i-n diode DetectorPillars [6], measured efficiencies of 4.5% for natural boron and 21%for 95% enriched 10B [7] and 35% measured thermal neutrondetection efficiency with the 6LiFcap sinusoidal trench perforateddevice was reported by [8]. Although in perforated detectors, thedetection efficiency is superior than in semiconductor planardetector, Si perforation requires special tools and equipment inorder to grow or carve the desired geometry in the Si substrate byVery Large Scale Integration (VLSI) or Etching techniques, and

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/nima

Nuclear Instruments and Methods inPhysics Research A

http://dx.doi.org/10.1016/j.nima.2014.04.0220168-9002/& 2014 Published by Elsevier B.V.

1 Q is the amount of energy released from the reaction.

Nuclear Instruments and Methods in Physics Research A 756 (2014) 62–67

homogenously fill these cavities with the neutron absorbingmaterial usually by Chemical Vapor Deposition (CVD).

Neutron detection based on the scintillation mechanism assuggested in model (B) is not introduced here for the first time.Boron loaded scintillators for neutron detection is a well-knowntechnology. It is worth mentioning that the Phoswich detector [9]and boron loaded elastomer coupled to gamma scintillator forneutron detection in oil logging applications [10] are also based onthe same physical detection mechanism. However, the innovationin this configuration is the usage of natural boron as a neutronabsorber in order to enhance the detection capabilities of scintil-lating gamma ray detectors in HLS application by measuring theneutrons emitted by Specific Nuclear Materials (SNM).

2. Modeling and experimental methods

2.1. Monte-Carlo simulations

Two Monte-Carlo algorithms written in MATLAB code weredeveloped in order to optimize the efficiency for B4C thickness.Code (A) calculates the optimal thickness of boron carbide fordetecting the scattered charged particles from the 10B(n,α)7Lireaction. While code (B) calculates the thickness for detectingthe 478 keV photons emitted in 94% of the cases from the excitedLi. Both codes examined the boron carbide with natural atomicfraction of 10B in B (19.8 at%) and highly enriched 10B (96 at%) inB4C compounds. Fig. 1 presents a geometrical description of themodels, where in Fig. 1:

t—Thickness of Boron-Carbide coating (z axis),a—Width of the coating (x axis).b—Depth of the coating (y axis).A—Surface area of B4C (ab).z—Distance at which a thermal neutron interacts with 10Bnucleus.φα—Azimuth angle of scattered alpha particle.θα—Polar angle of scattered alpha particle.

In each “run” of the Monte-Carlo code, random numbers aregenerated under the following assumptions2:

a. All neutrons arriving to the detector are thermal (0.0253 eV)and perpendicular to the detector's surface A at uniformlydistributed entrance points:a0¼ a�m—x coordinate of neutron entrance to the boroncarbide coating.b0¼ b� n—y coordinate of neutron entrance to the boroncarbide coating.

b. Neutrons, charged particles and photons traverse in a trajectoryof a straight line in matter.

c. Distance z at which neutrons are absorbed in 10B is exponen-tially distributed according to calculated macroscopic cross-section.

d. The ratio of the absorption in boron-10 to all the otherpossibilities:

10Bratio ¼NB10sa_B10

Σtot_B4C

The occurrence will be determined with the uniformly dis-tributed random number d.

(NB10—atomic density of 10B in the compound, sa_B10—micro-scopic cross-section of absorption in 10B for thermal neutrons,Σtot_B4C—total macroscopic cross-section for thermal neutrons).Cross-sections for neutrons were extracted from the ENDF/B-VII.0 cross-sections library from the java based nuclear data-base software: Janis 3.4 [11].

e. The reaction 10B(n,α)7Li have two possible results:105 Bþn-7

3Liþα—6% of the reactions;105 Bþn-7

3Linþαþγð0:48 MeVÞ—94% of the reactions.

From energy and momentum conservations laws:

ELi ¼ 1:01 MeV ; Eα ¼ 1:78 MeV; ð6%ÞELi ¼ 0:84 MeV ; Eα ¼ 1:47 MeV ; ð94%Þ

The random number γ defines the chance of the occurringevent.

f. The charged particles are scattered in opposite directions(back-to-back). Alpha particle's (or photon's) solid angleΩαðθα;φαÞ is uniformly distributed. In case θα4901, the alphaparticle is scattered backwards and the lithium ion approachesthe detector with a solid angle ΩLið1800�θα;180þφαÞ, where:φα ¼ 2πp; θα ¼ cos �1ð1�2qÞ.

g. Gamma photons emission is isotropic. The Gamma photonchance of escaping the coating is exponentially distributed.Gamma photoelectric absorption coefficients are importedfrom the National Institute of Standards and Technology (NIST)web database for photons cross-sections, XCOM [12].

h. All calculations were conducted for 2� 2 cm2 ða� bÞsurface area.

i. Particles and photons which escape the geometry (from thesides) or absorbed during transition are counted as escapedparticles.

According to these assumptions, the interaction coordinatesare ða0; b0; zÞ, and the exit coordinates are ðXX;YY ; tÞ where: XX ¼a0þR sin ðθÞ cos ðφÞ—x coordinate of the charged particle exitpoint. And, YY ¼ b0þR sin ðθÞ sin ðφÞ—y coordinate of the chargedparticle exit point.

The stopping powers ðdE=dxÞ for the charged particles emittedin the neutron–boron reaction were evaluated using theStopping and Range of Ions in Matter (SRIM) software [13] tablesof boron and carbon and scaled using the Bragg–Kleeman rule forcompounds.

t

xLi

αt-x

ϕ

θ

Incident thermal neutron Surface Area: A

Detector xy z

ab

Fig. 1. Geometrical description for the Monte-Carlo based codes.

2 p,q,d,m,n,γ are uniformly distributed random numbers from 0 to 1.

Y. Ben-Galim et al. / Nuclear Instruments and Methods in Physics Research A 756 (2014) 62–67 63

The distance R from the neutron's interaction point to thedetector's surface is evaluated as shown in Eq. (1):

R¼ t�z= cos ðθÞ ð1Þ

The energy of ions arrived to the detector's surface (after distanceR in the compound) is a numeric solution (with a resolution of100 eV) of the higher limit of the integral expressed in Eq. (2):

R¼ �Z Ef

Ei

dEðdE=dxÞ ð2Þ

The ions energy has to be detected in order for the neutron tobe counted.

Fig. 2 presents the scaled stopping powers for alpha andlithium ions in B4C for natural and enriched boron.

The difference in stopping power between natural andenriched boron in the B4C compound is caused by differencein their densities and average atomic masses (Atomic MassUnit—amu).

Plots of the remaining energy of the ions over the rangecalculated from Eq. (2) are shown in Fig. 3.

The probability of gamma reaching the detector's surface isexpressed and evaluated by Eq. (3):

P ¼ e�μB4CR ð3Þ

where, μB4C is the gamma linear photoelectric absorption coeffi-cient in the compounds.

For both codes the Monte-Carlo calculations are shown by thealgorithms given in Fig. 4.

2.2. Materials and set-up analysis

The equipment used for the experiment is based on a CsIscintillator detector with dimensions of 3 in. diameter and heightof 3 in. coupled with a boron-carbide disc. The disc is B4Csputtering target material supplied by PLASMATERIALS (USA) with99.5% purity (composed of natural boron which contain 19.8% 10B)and dimensions of 2 in. diameter and thickness of 0.125 in. Theapplied equipment and calibration setting used for the experimentare as follows: High voltage supply: set to 603 V; Pulse shapingamplifier with Shaping time of 2 μs and a multi-channel Spectrumanalyzer. The experiment configuration is presented in Fig. 5. Thescintillator detector was calibrated using 57Co (122 keV) and 60Co(1174 keV, 1332 keV) sources. The keV/Channel linear calibrationfit was

E½keV� ¼ 0:753 N�3 ð4ÞThe detector FWHM (Full width at half maximum) resolutionfor 137Cs (662 keV) is about 8% and nearly 10% for 478 keVphotons. In order to increase the measuring sensitivity in orderto capture Compton scattered photons, the Region of interest (ROI)energy window was set to 78% from the expected 478 keV photo-peak. The Measuring time was set to 100 s

The total measured efficiency will be evaluated from thefollowing expression:

ηtot ¼CPS

AGϕratioð5Þ

where in Eq. (5): ηtot—Total efficiency of the detector, CPS—Number of neutron counts per second., A—Neutron activity ofthe source (number of neutrons per second)., G – Geometry factor– determined for the ratio the detector and source areas, ϕratio—

Fraction of thermalization of neutrons.The neutron source used in the experiment was a 252Cf with

neutron activity of �10,000 Bq at the time the experiments wereheld. The geometrical factor was calculated: G¼13% (the ratio

104 105 1060

2

4

6

8

10

Lithium/Alpha Ion Energy [eV]

dE/d

x [e

V/m

]

Lithium, %atomic10B=96%

Lithium, %atomic10B=19.8%

Alpha, %atomic10B=96%

Alpha, %atomic10B=19.8%

x 1011

Fig. 2. Alpha and lithium particles stopping powers in B4C compound for naturaland enriched boron.

0 1 2 3 4 50

0.5

1

1.5

]Ve

M[ygren

Egninia

meR

noI

Ealpha1=1.47[MeV], %atomic10B=19.8%

Ealpha1=1.47[MeV], %atomic10B=96%

Elithium1=0.84[MeV], %atomic10B=19.8%

Elithium1=0.84[MeV], %atomic10B=96%

0 1 2 3 4 50

0.5

1

1.5

Range [μm] Range [μm]

]Ve

M[ygren

Egninia

meR

noI

Ealpha2=1.78[MeV], %atomic10B=19.8%

Ealpha2=1.78[MeV], %atomic10B=96%

Elithium2=1.01[MeV], %atomic10B=19.8%

Elithium2=1.01[MeV], %atomic10B=96%

Fig. 3. Remaining ion's energy over range. (a) 6% Of the interacions. (b) 94% of the interactions.

Y. Ben-Galim et al. / Nuclear Instruments and Methods in Physics Research A 756 (2014) 62–6764

between the solid angle Ω of a cone which is base is the circularbase of the B4C disc and the total solid angle of the isotropicemission of the source—4π). Beyond 3 cm of polyethylene mod-eration [14], the population of thermal neutrons (at r10–5 eV)

increases to about 6% of the total neutron flux. At higher thickness,the moderation rate is similar to the absorption rate, so thethermal neutron population does not increase. The same can beassumed for polypropylene (PP) since both PP and polyethylenehave similar moderating properties. Hence, for 1 in. of PP, thethermalization ratio: ϕratio � 0:06.

3. Results and discussion

3.1. Simulations results

The results obtained from the algorithm shown in Fig. 4(a) forions capture are presented in Fig. 6. Fig. 6(a) shows the ITNE atdifferent coating thicknesses for ions detection, while Fig. 6(b) showsthe spectra (without noise) at thickness of maximum ITNE.

The two visible peaks on the spectra plotted in Fig. 6(b) indicatethe energy peaks of lithium and alpha particles occurring in 94% ofthe 10B(n,α)7Li interactions. These peaks are detected when theneutron interacts with the boron-10 in the vicinity of the chargedparticles detector, while the ions partly lose their energy. For both10B compositions, as shown in the figure, the higher energy peakbelongs to the alpha particle (1.47 MeV) and the lower one isattributed to the lithium particle (0.84 MeV).

The optimized efficiency decreases in case Lower Level Dis-criminator (LLD) which is considered [3]. Furthermore, thesedetectors have thick entrance window where the incident parti-cles lose additional energy (increases LLD). The He and Li ionsemitted in the neutron–10B reaction, have lower energies

Fig. 4. Flow diagrams of suggested algorithms. (a) Ions capture. (b) Gamma capture.

Fig. 5. The experiment set-up configuration.

Y. Ben-Galim et al. / Nuclear Instruments and Methods in Physics Research A 756 (2014) 62–67 65

compared to natural alpha particles (e.g. alpha emission fromRadon 45 MeV). The absorbed energy in the entrance windowof silicon solid state detectors, significantly limits the detectionefficiency of the neutron reaction particles. Thus, in order toimprove the detection efficiency, ion implanted detectors withthin entrance window are recommended. The detection area of theion implanted detectors is small, so their applications for neutrondetection in HLS are limited for long time measurements. Thislimitation does not affect the fast detection of neutron emittingmaterials. Therefore, it can be concluded that the efficiency ofthermal neutrons detection in continuous alpha counting siliconsolid state detectors coated by natural boron B4C compound isabout 1%. By using a B4C layer with enriched 10B (see Fig. 6(a)), thethermal neutrons detection efficiency increases to �5%. This valuewas found in agreement to that reported in Section 1. Thissensitivity limits the performance of neutron detection. For photondetection, the simulation results are more promising. Fig. 7 pre-sents the ITNE obtained for gamma-photon detection at variousB4C coating thicknesses:

It can be seen in Fig. 7 that a significant ITNE value (40–44)% wasobserved for detecting gamma photons at thickness of �120 μmusing enriched boron, while for natural boron compound, that

thickness observed at �470 μm. This means that efficient neutrondetection is feasible using natural boron compounds. The overallneutron detection efficiency may be affected by the efficiency of thegamma scintillator detector, which simulations and empiric testsshows an intrinsic efficiency greater than 30% for 478 keV photons,for a typical 3 in.�3 in. CsI(Tl) or NaI(Tl) gamma detector. However,this detector configuration still maintains higher overall detectionefficiency compared with the calculated assessments of semiconduc-tor planar neutron detectors. The drawback of this configuration isgamma sensitivity. In order to improve the rejection of other 478 keVphotons due to Compton interactions, a pair of similar gammadetectors should be applied. In a pair detectors system, one shouldbe coupled to a B4C coating for Gamma-neutron measurement andthe second for the Gamma rejection. This configuration may beuseful for the large scale applications such as RPMs.

3.2. Experimental data analysis

Fig. 8 shows that the optimal ITNE measured using the CsIscintillator detector with a B4C disc (2 in. diameter�0.125 in.thick) is �41.5%.

100 101 102 103 104 105 1060

1000

2000

3000

4000

Ion Energy [eV]

Rel

ativ

e C

ount

s

Spectrum for %atomic10B=96%

Spectrum for %atomic10B=19.8%

α

Li

αLi

0 5 10 150

1

2

3

4

5

Depth of BC coating [μm]

Effi

cien

cy [%

]

%atomic10B=96%

%atomic10B=19.8%

(3.1 ,4.7%)m μ

(3.4 ,1.05%)μm

Fig. 6. (a) Thermal efficiency of the coating for ions detection. (b) Counting Spectra at maximum efficiency coating thickness.

0 200 400 600 8000

10

20

30

40

50

Depth of BC coating [μm]

Effi

cien

cy [%

]

%atomic10B=96%

%atomic10B=19.8%

(470 ,40%)mμ

(120 ,44.6%)mμ

Fig. 7. Thermal efficiency of gamma photons detection as function of coatingthickness.

(3175 ,41.5%)mμ

Fig. 8. Intrinsic thermal neutron efficiency over thickness of natural B4C discdesigned for the experimental model.

Y. Ben-Galim et al. / Nuclear Instruments and Methods in Physics Research A 756 (2014) 62–6766

In order to assess the total efficiency of the detector, threeadjustments to the mathematical model were performed: (i) asection which calculates the photo-electric intrinsic efficiency inthe scintillator and in air; (ii) conversion to calculation in cylind-rical coordinates; and (iii) corrections which were needed to beaccomplished since the diameters of the scintillator and B4C discwere inconsistent. The total efficiency obtained from the MATLABcode yielded 10%. In addition, an MCNP simulation (VersionMCNP4C2 [15]) for this configuration yielded probability of �9%that a photon will be absorbed in the CsI scintillator. These resultswere found to be consistence to the �41.5% ITNE of the B4C layermultiplied by the 30% photo-electric intrinsic efficiency of thescintillator and additional loses due to photon absorption in theB4C disc.

Fig. 9 shows two spectra measured by using the experimentalconfiguration given in Fig. 5. Fig. 9(a) shows the spectrumobtained from the scintillator positioned in front of the 252Cfsource, while Fig. 9(b) shows that spectrumwhen the B4C disc wasadded as a neutron absorber.

The counts in the ROI for the spectrum (a) is 1435 and for thatof spectrum (b) is 2406. The difference between the two spectrais 971 counts. This value indicates on the amount of neutronsabsorbed in the 10B isotope, as result of a photon emissiondetected by the scintillator. Hence, for measuring time of 100 s,CPS equals 9.71 neutron/s and the total efficiency measured forthis experiment (according to Eq. (5)) is �12.4%.

4. Conclusions

This study suggests a new approach to enhance neutrondetection capabilities, which is feasible, easy to assemble and

relatively inexpensive. The detection efficiency can be obtained bymeasuring the 478 keV photons that are emitted in 94% of the 10B(n,α)7Li interactions. Experimental measurements for the pro-posed detector yielded efficiency value of 12.4%. Even though thisvalue is not the highest observed, this suggests that a naturalboron compound based detector on gamma detection technologyis feasible in the nuclear, medical and HLS industries. The analysisfor various applications where gamma detectors are widely avail-able (e.g. HLS) can be performed by subtracting the signals fromtwo similar scintillator detectors; one is a reference, and thesecond is coupled with a B4C disc. For other applications, it isfeasible to develop an electronic circuit that can collect only thedesired photo-peak counts.

Acknowledgements

The authors would like to thanks Mr. Dan Peer (CEO of RotemIndustries Ltd.) for his encouragement and his support for thisproject. This research was supported by a joint grant from RotemIndustries Ltd. and the Israel Atomic Energy Commission forInternational Collaborations (Project no. 2013-50183) in theresearch of HLS detectors.

References

[1] R.T. Kouzes, J.H. Ely, Pacific Northwest National Laboratory (2010).[2] B.J. Allen, Neutrons in cancer therapy, in: International Conference on

Neutrons and Their Applications 494 (March 3, 1995), http://dx.doi.org/10.1117/12.204193, 1995.

[3] D. McGregor, M. Hammig, Y. Yang, H. Gersch, R. Klannd, Nuclear Instrumentsand Methods Physics Research 1-3 (500) (2003) 272.

[4] Z. Wang, C.L. Morris, Nuclear Instruments and Methods Physics ResearchSection A (652) (2011) 323.

[5] J.K. Shultis, D.S. Mcgregor, Efficiencies of coated and perforated semiconductorneutron detectors, in: Proceedings of the Nuclear Science Symposium Con-ference Record, IEEE, Vol. 7, pp. 4569–4574.

[6] R.J. Nikolic, C.L. Cheung, C.E. Reinhardt, T.F. Wang, Future of SemiconductorBased Thermal Neutron Detectors, Barry Chin Li Cheung Publications, Boston,MA, United States, 2006 (February 24).

[7] Y. Danon, J. Clinton, K.C. Huang, N. LiCausi, R. Dahal, J.J.Q. Lu, I. Bhat, Journal ofInstrumentation 7 (2012).

[8] S. Bellinger, W. McNeil, T. Unruh, D. McGregor, Angular response of perforatedsilicon diode high efficiency neutron detectors, in: Nuclear Science Sympo-sium Conference Record, NSS ’07, Vo. 7 IEEE, Honolulu, HI.

[9] M.C. Miller, R.S. Biddle, S.C. Bourret, R.C. Byrd, N. Ensslin, W.C. Feldman, J.J. Kuropatwinski, J.L. Longmire, M.S. Krick, D.R. Mayo, P.A. Russo, M.R. Sweet,Nuclear Instruments and Methods in Physics Research Section A: Accelerators,Spectrometers, Detectors and Associated Equipment 422 (1–3) (1999) 89, ISSN0168-9002, http://dx.doi.org/10.1016/S0168-9002(98)01069-9.

[10] D.G. Penn, Thermal Neutron Detector Using a Scintillator with BackgroundGamma Ray Shielding. United States Patent US006373066B1, 16 April 2002.

[11] (ENDF/B-VII.0), OECD Nuclear Energy Agency—JANIS 3.4. [online]. Available:⟨https://www.oecd-nea.org/janis/⟩.

[12] XCOM – Photon Cross-Sections Database, NIST – National Institute of Stan-dards and Technology's, [online]. Available: ⟨http://www.nist.gov/pml/data/xcom/index.cfm⟩.

[13] [online]. Available: ⟨http://www.SRIM.org⟩.[14] R. Runkle, A. Bernstein, P. Vanier, Securing Special Nuclear Material: Recent

Advances in Neutron Detection and Their Role in Nonproliferation Vol. 108(11) (2010).

[15] Monte Carlo N-Particle Transport Code System, version 4C2, Technical ReportLA-13 709-M, Los Alamos National Laboratory, NM, USA, 2001.

Channel

Cou

nts

Cou

nts

Fig. 9. Spectrum of intensity vs. energy. (a) The scintillator positioned in front ofthe 252Cf source. (b) B4C disc was added as a neutron absorber.

Y. Ben-Galim et al. / Nuclear Instruments and Methods in Physics Research A 756 (2014) 62–67 67