the umd/nist fast neutron spectrometers

The UMD/NIST Fast Neutron Spectrometers

T.J. Langford Yale University

E.J. Beise, H. Breuer University of Maryland

C.R. Heimbach, J.S. Nico National Institute of Standards and Technology

��1

Outline• Fast neutrons for low-background science

• At sea-level

• Underground

• Previous detection techniques

• Capture gated spectroscopy

• The FaNS detectors

• FaNS-1: The proof-of-principle

• FaNS-2: The purpose-built detector

• Outlook for FaNS

T.J. Langford WIDG Seminar 2/18/2014��2

Fast Neutrons for Low-Background Physics


Fast Neutrons for Low Background Science

• Fast neutrons play a particularly problematic role in low background experiments

• Deeply penetrating, and difficult to shield

• Surface → Activation of shielding and detector materials

• Backgrounds for experiments that must run sea-level

• Underground → FNs are indistinguishable from WIMP dark matter interactions

• Important to know the fast neutron spectra in both environments for experiment design and background rejection


Sea-level Cosmogenic Neutrons• Very high energy neutrons are

created in cosmic ray air showers

• Low background materials are required for many experiments, especially (Ge, Pb, Cu)

• Determination of material activation requires precision knowledge of neutron flux and spectrum

• Large variations among measurements and simulations

• Large fluctuations with local environment and conditions

T.J. Langford WIDG Seminar 2/18/2014

3432 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 51, NO. 6, DECEMBER 2004

TABLE IIFITTED PARAMETERS FOR THE ANALYTIC MODEL

. In the Appendix, we provide a table of numer-ical values of neutron differential flux, , fordetermined from the shape of the Yorktown Heights spectrumscaled down to the location of New York City (NYC) at sealevel (as it is shown in Fig. 6), then further adjusted up to the fitin Fig. 5, and up to the mid-point of solar modulation using (3).We refer to this spectrum as the “reference” spectrum. Its flux is0.901 times that of the measured spectrum at Yorktown Heightsand 1.07 times that of the scaled spectrum shown in Fig. 6.

The spectrum may change shape slightly in the GeV region athigher cutoffs. The airplane measurements of Goldhagen et al.[14], obtained at an altitude of 20 km, found that the fraction ofthe total flux that was above 10 MeV was 8% higher at a cutoffof 11.8 GV than at 0.8 GV. Calculations by Mares et al. [37]show that the cutoff dependence of the spectrum shape is lessthan half as much on the ground as it is at 20 km.

V. ANALYTIC MODEL

In addition to tabulated values of , a simple analyticexpression has been fit to the NYC reference neutron spectrumin the energy range from 0.1 MeV to 10 GeV

(6)

The values of the parameters , , and in (6) wereconstrained by requiring that the energy-integrated fluence,

, in the energy regions of the evaporation andhigh-energy peaks from the model agree with the experimentaldata to within a few percent. The numerical values of theseparameters are listed in Table II. Since the flux in the thermalpeak, and to a lesser extent the plateau region, depend onthe local environment, functions fitting these regions are notpresented here.

Fig. 7 shows a graph of the evaporation and high-energy re-gions of the reference spectrum and the analytic model in the

representation. The data are shown as a histogramand the analytic model is the solid smooth curve. The fit is vis-ibly very good above 10 MeV and reasonably good in the evap-oration region down to about 0.4 MeV.

VI. COMPARISON TO JEDEC STANDARD

Fig. 7 also shows the spectrum from Appendix E of JEDECStandard JESD89 [38]. The JEDEC model was a fit throughpreviously published data adjusted to the same conditions asour reference spectrum. (This was the main reason we chosethose conditions.) It has been used for a number of years andforms the basis of current SER calculations. The JEDEC modelunderestimates the reference measured flux integrated from 50MeV to 1 GeV and overestimates

Fig. 7. Upper-energy portion of the reference neutron spectrum at New YorkCity, sea level, and mid-level solar modulation (histogram), the analytic fit (solidsmooth curve), and the model from Appendix E of JEDEC Standard no. 89 [38](dashed curve).

Fig. 8. Differential flux, , of cosmic-ray inducedneutrons as a function of neutron energy. The data points are our referencespectrum from the measurements, the solid curve is our analytic model, andthe dashed curve is the JEDEC model [38].

it from 5 to 50 MeV and again from 1 to 10 GeV by factors of1.7 and 1.3, respectively.

Fig. 8 shows the same data as Fig. 7, but presented as themore familiar differential flux (neutrons ).As already shown in Fig. 7, our analytic model reproduces themeasured spectrum better than the JEDEC model.

VII. CONCLUSION

Five sets of neutron spectrometer data have been collectedand analyzed from a variety of sites across the United Statesto determine the flux and energy distribution of cosmic-ray in-duced neutrons on the ground. The measurement sites had awide range of altitudes and a small range of geomagnetic cutoffrigidities (1.6 to 4.7 GV). An extended-energy Bonner spherespectrometer was used that collected data simultaneously acrossan energy range from 1 meV to about 10 GeV. The measuredtotal neutron flux varied by a factor of 15 from the highestto lowest altitude sites, but the shape of the spectrum above

Gordon et al. (2004) FN Spectrum with 14

Bonner Spheres

��5

Sea-level Cosmogenic Neutrons• Very high energy neutrons are

created in cosmic ray air showers

• Low background materials are required for many experiments, especially (Ge, Pb, Cu)

• Determination of material activation requires precision knowledge of neutron flux and spectrum

• Large variations among measurements and simulations

• Large fluctuations with local environment and conditions


- 3 measurements with the same detector

- 3 very different fluxes

0.0012

0.0010

0.0008

0.0006

0.0004

0.0002

0.0000

(E x

Flu

x) (

n/cm

2 /s)

101 102 103 104

Energy (MeV)

Ziegler 1996 Gordon 2004 CRY Simulation

3 different FN spectra

��6Nakamura et al. 2005

Underground Neutrons


• 1-10MeV: Natural radioactivity from surround material, mainly (alpha,n) (location dependent)

• 1 MeV - GeV: Muon-induced spallation neutrons (depth dependent)

• Addition of shielding materials (like Pb) can enhance the production of muon-induced neutrons

15

events/keV-kg-year, is precisely what we have found inour simulation.

Depth (km.w.e.)1 2 3 4 5 6

)-1 y

ear

-1 k

g-1

Even

ts (k

eV

-710

-610

-510

-410

-310

-210

-110

12510

2610

2710

2810

Hal

f Life

(yea

rs)

Majorana

Granularity, PSD and Segmentation

Target sensitivity

Without neutron veto

With neutron veto (99%)

WIP

P

Gra

n Sa

sso

Sudb

ury

FIG. 25: The Depth-Sensitivity-Relation (DSR) derived fora Majorana-like experiment showing, specifically, the resultsfrom this work assuming the detector is operated at a depthequivalent to the Gran Sasso Laboratory. The raw event ratein the energy region of interest of 0.026 events/keV-kg-yearcan be reduced by a factor of 7.4 by exploiting the detectorgranularity, pulse-shape discrimination (PSD), and detectorsegmentation. The upper curve displays the background sim-ulated in the case that no active neutron veto is present andthe lower curve indicates the reduction that would ensue if anactive neutron veto were present that is 99% efficient.

The results of our simulations can be used to derivethe DSR for Majorana as shown in Fig. 25. The neu-tron induced background can be reduced by about a fac-tor of 7.4 in Majorana owing to the use of crystal-to-crystal coincidences and the use of pulse-shape discrim-ination and segmentation. Nonetheless, to achieve thetarget sensitivity of next generation double-beta decayexperiments, 0.00025 events/keV-kg-year correspondingto the background level required to reach sensitivity tothe atmospheric mass scale of 45 meV Majorana neutrinomass, the muon-induced background must be reduced byroughly another factor of 100. This can be achieved onlyby operating such a detector at depths in excess of 5km.w.e., otherwise an active neutron veto would need tobe implemented with an efficiency in excess of 99%.

D. (α,n) Background

Once the depth requirement is satisfied, a proper shieldagainst (α,n) neutrons from the environment becomesnecessary. We use the standard rock and the measuredneutron flux (3.78×10−6cm−2s−1 [68, 69] ) at Gran Sassoassuming that all underground labs have the same orderof neutron flux to establish the shielding requirement for(α, n) neutrons. This flux corresponds to an average ofabout 2.63 ppm 238U and 0.74 ppm 232Th activity inGran Sasso rock and 1.05 ppm 238U and 0.67 ppm 232Thactivity in Gran Sasso concrete [70]. The neutron energy

spectrum depending on the rock composition is shown inFig. 26. As can be seen, the total neutron flux is aboutthree orders of magnitude higher than that of neutronsfrom the rock due to muon-induced processes, but theenergy spectrum is much softer.

Neutron Energy (MeV)0 1 2 3 4 5 6 7 8 9 10

)-1 M

eV-1

s-2

Neu

tron

s (c

m

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

0.2

0.22-610×

Standard Rock

Gran Sasso Rock

FIG. 26: The neutron energy spectrum arising from (α,n)reactions due to radioactivity in the rock. We predict a harderenergy spectrum in Gran Sasso rock relative to standard rockowing to the presence of carbon and magnesium.

To demonstrate the neutron flux and energy spectrumat different boundaries we show the rock/cavern neutronflux and energy spectrum with a shielding for Majoranadescribed earlier in Fig. 27 for the depth of Gran Sassoas an example.

Energy (MeV)-110 1 10 210 310

)-1 M

eV-1

y-2

Neu

tron

s (c

m

-1210-1110

-1010

-910

-810

-710

-610

-510

-410

-310

-210

-1101

10

210

310:Rock/Cavern boundaryµ

:After poly shieldingµ

:After lead + copper shieldingµ

,n): Rock/Cavern boundaryα(

,n): After poly + lead + copper shieldingα(

FIG. 27: The energy spectrum for fast neutrons producedby (α, n) reactions in the rock compared to those induced bymuon interactions in the rock with and without shielding. Thelower energy neutrons (< 10 MeV) are quickly absorbed usingpolyethylene shielding, however, the high energy portion ofthe muon-induced neutron flux persists. The addition of leadshielding adjacent to a detector can also create an additionalsource of muon-induced neutrons.

Note that the (α, n) neutrons from the rock are quicklyattenuated to the level of the muon-induced neutrons be-

(↵, n)

8

the corrected neutron multiplicity (equation (10)) andthe muon fluxes and distributions outlined in Section II.The neutron flux (φn) as a function of depth is shownin Fig. 14 where we have included a fit function of thefollowing form:

φn = P0(P1

h0)e−h0/P1 , (13)

where h0 is the equivalent vertical depth (in km.w.e.)relative to a flat overburden. The fit parameters are P0 =(4.0 ± 1.1)×10−7 cm−2s−1 and P1 = 0.86 ± 0.05 km.w.e..

Depth (km.w.e.)2 3 4 5 6

)-1 s

-2N

eutr

on F

lux

(cm

-1110

-1010

-910

-810

WIPPSoudan

Kamioka

Boulby Gran Sasso

Sudbury

FIG. 14: The total muon-induced neutron flux deduced for

the various underground sites displayed. Uncertainties on

each point reflect those added in quadrature from uncertain-

ties in knowledge of the absolute muon fluxes and neutron

production rates based upon our simulations constrained by

the available experimental data.

In Table V we summarize the neutron flux at therock/cavern boundary for the various sites consideredand note that we have not included the effect of neutronsthat emerge from one surface and back-scatter back intothe cavity. The results are in good agreement with theexisting simulation results for Gran Sasso [42]. If thesimulation results for Boulby [65] are modified using ourneutron multiplicity correction, good agreement is alsofound between the two results. It is relevant to note thatthere is a significant fraction of the neutrons with energyabove 10 MeV.

TABLE V: The muon-induced neutron flux for six sites (in

units of 10−9 cm−2s−1). The total flux is included along with

those predicted for neutron energies above 1, 10, and 100

MeV.

Site total > 1.0 MeV > 10MeV > 100MeV

WIPP 34.1 10.78 7.51 1.557Soudan 16.9 5.84 4.73 1.073Kamioka 12.3 3.82 3.24 0.813Boulby 4.86 1.34 1.11 0.277Gran Sasso 2.72 0.81 0.73 0.201Sudbury 0.054 0.020 0.018 0.005

2. Neutron Production in Common Shielding Materials

Fast neutrons can also be created by muons passingthrough the materials commonly used to shield a detectortarget from natural radioactivity local to the surroundingcavern rock. Fig. 15 shows the neutron yield in somecommon shielding materials. We have also included asimulation for germanium which will prove useful laterin this paper when we consider the DSR for experimentsbased on this target material.

Depth (km.w.e.)2 3 4 5 6

)-1

s-3

Neu

tron

Pro

duct

ion

Rat

e (n

cm

-1310

-1210

-1110

-1010

-910

-810 WIPPSoudan

KamiokaBoulby

Gran Sasso

Sudbury

LeadPolyethyleneCopperGermanium

FIG. 15: The muon-induced neutron production rate pre-

dicted for some common detector shielding materials. Note

that minor variations due to neutron back-scattering have

been neglected in these calculations.

The fitted functions have the same form as equa-tion (13) but with different values for parameters whichare provided in Table VI. To convert the neutron produc-tion rate to the total neutron flux, one multiplies equa-tion (13) by the average muon path length which dependsupon the detector geometry.

TABLE VI: Summary of the fitting parameters describing the

muon-induced neutron production rate in common detector

shielding materials.

Material P0 P1

Lead (7.84±2.21) × 10−8 0.86±0.05

Polyethylene (6.89±1.95) × 10−9 0.86±0.05

Copper (2.97±0.838) × 10−8 0.87±0.05

Germanium (3.35±0.95) × 10−8 0.87±0.05

Generally speaking, muon-induced neutrons producedin a detector target or surrounding shield can be activelyvetoed in coincidence with the primary, depending on theveto efficiency and specific detector geometry. Specificexamples are provided later in this paper.

Energy (MeV)-110 1 10 210 310

)-1 M

eV-1

y-2

Neu

tron

s (c

m

-1210-1110

-1010

-910

-810

-710

-610

-510

-410

-310

-210

-1101

10

210






3

N’

µ µ ’

γ

n,π

N

(a)

γ

n,π

NN’

(b)

FIG. 1: Feynman diagrams depicting muon spallation (a) and photo-absoprtion (b). In panel (a)

an incoming muon exchanges a virtual photon with a nucleus. Panel (b) shows a real photon being

absorbed by the nucleus. The nucleus subsequently emits neutrons or pions.

the e↵ects of secondary processes, such as neutrons produced via primary pions or primary

neutrons.

II. MUON SPALLATION THEORY

In muon spallation, an incoming muon interacts with a nucleus via an exchange of a

virtual photon. The nucleus subsequently emits neutrons (or pions). Figure 1(a) shows a

Feynman diagram depicting this scenario. The di↵erential cross section for the inclusive

reaction of a charged lepton (see, for example, Ref. [7]), such as the muon, is

d�

dEf

d⌦f

=↵2

q4

pf

pi

Lµ⌫Wµ⌫

, (1)

where ↵ is the fine structure constant, pi

and pf

are the initial and final lepton momenta,

respectively, q is the four-momentum transfer, Lµ⌫ is the lepton tensor, and W µ⌫ is the

hardronic structure tensor. The lepton tensor is known analytically,

Lµ⌫ = 2⇥pµ

i

p⌫

f

+ p⌫

i

pµ

f

+ gµ⌫(m2 � pi

· pf

)⇤, (2)

where m is the lepton mass. The hadronic tensor, on the other hand, depends on nuclear

details that are not know analytically and must be empirically parametrized. It does satisfy

certain symmetries, however, which constrains the parametrization. For example, it must

��7Plots from Mei and Hime 2006

Neutron Production Simulation vs Data

Finally, it is apparent that the experimental pointslie far above the MC data. If the disagreement forthe light materials is already unreasonable, thecase of lead cannot be easily explained. In fact,even if a much lower neutron threshold of 1MeVis considered, both MC models would still predicta lower cross-section than those reported inRef. [18].

The production of lower energy neutrons bydeep inelastic scattering (DIS) of 470GeV muonsin lead was studied by the E665 Collaboration,who found average neutron multiplicities per DISevent of ’ 5 for neutron energies under 10 MeV[19]. A value of 3.7 is obtained for the GEANT4simulation of the m–N process at this muon energy,in reasonable agreement with the experimentalresult. The simulated neutron spectrum (per unitenergy) exhibits a double-exponential behaviourbelow 10MeV—also in agreement with the experi-mental findings. The two decay constants, due toneutron evaporation from the thermalised nucleusand from pre-equilibrium emission, are charac-

terised by nuclear temperatures of 0.93 and3.7MeV, compared to 0:7! 0:05 and 5! 1MeVobtained in E665. In conclusion, the spallation ofneutrons under 10MeV as predicted by GEANT4for lead does not conflict with these experimentaldata.The role of the minimum energy transfer in the

muon photonuclear models in neutron productionwas pointed out in Paper 2. This threshold comesabout because the virtuality of the photon can nolonger be neglected when it becomes comparableto its energy. Recently, the total m–N cross-sectionwas reported to increase by 2–3 times if theminimum energy transfer is decreased from 140 to10MeV, based on the parameterisation used inFLUKA [20]. We have confirmed that thisdifference is only 10–15% greater for the200MeV threshold in GEANT4. The aforemen-tioned study also found that the parameterisationused to describe the g–N cross-section in FLUKA[16] (similar to that from Ref. [17] used inGEANT4) overestimates more rigorous theoreti-cal calculations when extrapolated to low energygammas. Consequently, the increase in the muoncross-section with decreasing threshold is notexpected to be as large as mentioned above. Inany case, as pointed out in Paper 2, we expectmany more neutrons to be produced by brems-strahlung (real) photons with low energies inelectromagnetic cascades than by virtual ones inmuon interactions with small energy transfers.

3. Underground neutron fluxes: a case study

The UK Dark Matter Collaboration (UKDMC)has been assessing the feasibility of a xenon-basedtonne-scale dark matter experiment to be installedat the Boulby Underground Laboratory. In thiscontext, initial calculations using FLUKA, re-ported in Paper 3, have so far been performed ofthe muon-induced background in a 250 kg xenontarget. Building on that work we present here acase-study comparison between FLUKA andGEANT4. The calculated neutron fluxes andspectra at the rock/cavern boundary and aftervarious shields are also relevant to other under-ground experiments in different laboratories.

ARTICLE IN PRESS

10-5

10-4

10-3

10-2

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1cos θ

dσ/dΩ

, bar

n/sr

CCuPb

Fig. 7. Differential cross-section of neutron production by190GeV muons for a 10 MeV threshold in neutron energy. Thedata points represent the results of the NA55 experiment. Thethin-line histogram shows the GEANT4 simulation consideringmuon–nucleus interaction only; the thick histogram includes allphysics processes. The dashed line represents the FLUKAresults for the latter case.

H.M. Araujo et al. / Nuclear Instruments and Methods in Physics Research A 545 (2005) 398–411 405

Data Simulation

n-production is severely

underestimated

Araújo, et. al. NIM A, 2005


Previous Detection Techniques


Bonner Sphere Arrays

• Passively moderated 3He counters • Deployed in arrays of ~10 detectors

with different diameters • Convert count rates into spectrum

• Relies on response functions • No direct energy observation


The geometry and material composition of the 16 counter/moderator combinations was implemented in GEANT4 as realisti-cally as possible. The counter wall was modelled as a 0.5 mm thickstainless steel shell with a 1.5 mm steel ring in the equatorial plane,a 9.25 cm long steel tube at the bottom (1.27 cm diameter, 0.5 mmthick wall) and a 0.95 cm long steel cap on top (1.14 cm diameter,0.5 mm thick wall). The 3He atom density inside the sphericalproportional counter was taken to be 4.25 ! 1019 cm"3. As anexample, Fig.1 shows the geometry of the 7 inch PE spherewith the3He counter in the center as used in the GEANT4 calculations.The density of the PE moderators was equal to 0.95 g cm"3 and thedensity of the leadwas equal to 11.337 g cm"3. The response of eachBonner sphere was calculated for monoenergetic neutrons(1 meVe10 GeV) impinging vertically on top of the sphere (seeFig. 1). The response data obtained from the MC calculations werethen interpolated to generate a response function with 130 energypoints in logarithmic equidistant intervals (i.e. 10 per decade) from1 meV to 10 GeV.

2.2. Bonner spheres spectrometer and unfolding

The data used to test the influence of the various calculatedresponse functions on the deconvoluted neutron fluence ratedistribution were obtained by means of a BSS system consisting of16 spherical 3He proportional counters (Ø 3.3 cm) mostly sensitiveto thermal neutrons. While one of these counters was operatedwithout any moderator, 13 others were surrounded by moderatorsincluding PE shells of various diameters (2.5, 3, 3.5, 4, 4.5, 5, 5.5, 6, 7,8, 9, 10, 11, 12 and 15 inches, respectively). Additionally, twocounters were surrounded by PE spheres of 9 inch diameter inwhich 0.5 inch and 1 inch thick lead shells were embedded. High-energy secondary neutrons from cosmic radiation, once moderatedand thermalized by these moderators, were finally detected in theproportional counters via the 3He(n,p)3H reaction. This system islocated at the Environmental Research Station (UFS) “Schnee-fernerhaus” (2650 m a.s.l.) on the Zugspitze mountain, Germany(longitude: E 10# 58,90, latitude: N 47# 250), within a measurement

shed constructed on a terrace of the UFS to house the spectrometer.The geomagnetic cutoff corresponded to 4.1 GV for the period ofJanuary and February 2008 (Bütikofer et al., 2007). A more detaileddescription of the geometry of each detector/moderator configu-ration and the measurement location is given in (Leuthold et al.,2007; Rühm et al., 2009a).

The system allows continuous measurements using all 16measurement channels simultaneously. For the present work,a measurement period of one month was chosen (February 2009),and the corresponding average readings of the 16 detectors duringthat period are shown in Fig. 2.

A spectral neutron fluence rate distribution was deduced fromthe experimental data shown in Fig. 2 using the MSANDB unfoldingcode (Matzke, 1987). The MSANDB code is a modified version of theSAND-II code (McElroy et al., 1967) that uses an iterative procedureand requires a priori physical information about the neutronspectrum (i.e. a start/guess spectrum), which represents a firstestimate of the shape of the neutron spectrum at the measurementlocation. Based on this input and the response matrix, MSANDBiteratively adjusts the neutron fluence rate distribution to beconsistent with the count rates measured by the various detectors.The relative uncertainties of the detector counts are used asweighting factors during the iteration process. After a certainnumber of iteration steps, the process is stopped. A guess spectrumincluding three peaks (at 25 meV, 1 MeV, and 100 MeV) and a flatepithermal region, and an iteration number of 300 were used forthe deconvolution process. For details see Simmer et al. (2010).

3. Results and discussion

3.1. Calculated response functions

Fig. 3 shows, as an example, the response functions of the 5, 8,and 12 inch spheres, as obtained using the different transport codesand INC models. It is obvious from the figure that some minordifferences occur in the energy range below about 20 MeV, whichresult from different cross section datasets and from the slightlydifferent geometries used in the calculations. However, significantdifferences are visible for energies above 20MeV. At these energies,the GEANT4 Bertini INC model clearly results in the highest valuesof the response functions, while the GEANT4 Binary INC modelappears to have the same energy dependence, but is consistentlylower. The values obtained with MCNP/LAHET and MCNP/HADRON

Fig. 1. Model of the 7 inch PE sphere with 3He counter as implemented in GEANT4.Tracks of neutrons hitting the sphere from above are also shown.

0

100

200

300

400

500

600

700

0 2 4 6 8 10 12 14 16

h[eta

RtnuoC

1-]

Sphere Diameter [inch]

Fig. 2. Count rates obtained in February 2009 at the UFS “Schneefernerhaus”, for the16 detector/moderator combinations used. X-axis denotes the diameter of the poly-ethylene (PE) spheres surrounding the detectors; 0 inch: detector was not surroundedby any PE; the increased count rates at 9 inch are due to 0.5 inch and 1 inch thick leadshells within the PE moderator.

C. Pioch et al. / Radiation Measurements 45 (2010) 1263e1267 1265

are also lowand roughly similar to those obtainedwith the GEANT4Binary INC model, up to an energy of about 1 GeV. For higherenergies, however, they decrease in contrast to those obtained withboth the GEANT4 Bertini INC and the GEANT4 Binary INC model.This trend is also seen for all other PE spheres (not shown).

For those PE spheres that contain lead shells, differencesbetween the various codes and models are observed in a similarway (Fig. 4) as for the pure PE spheres. The major differencesbecome obvious only above about 20 MeV. Again, use ofthe GEANT4 Bertini INC model provides the highest values for theresponse. In contrast to the results from the pure PE spheres,the MCNP/LAHET results for the spheres with lead are almost ashigh and expectedly do not decrease with increasing neutronenergy above 1 GeV. More specifically, they are very similar to thoseobtained with GEANT4, up to an energy of about 200 MeV. Abovethis energy the GEANT4 Binary INC provides values more thana factor of 2 lower than those obtained with the other codes. Notethat for these two spheres, no HADRON calculations were available,so the response functions of LAHET were used instead to determinefluence and dose rates.

The fact that for the pure PE spheres the GEANT4 Binary INCprovides higher values than the MCNP/LAHET codes while itprovides lower values for those PE spheres that include lead, mayindicate that the nuclear models used in the present work mightnot be equally suitable to model neutron interaction with light andheavy nuclei.

3.2. Neutron fluence rates and neutron ambient dose equivalentrates

While this finding is interesting for itself and asks for a moredetailed analysis of the nuclear models used, it must not necessarilymean that it is important in terms of unfolded neutron spectra, aswell as in terms of dose quantities. To research this inmoredetail, thecount rates shown in Fig. 2 were used and the spectral neutron flu-ence rate distributions were deconvoluted, using the four differentresponse function sets calculated as described above. The resultingfour neutron spectra are shown in Fig. 5 in the lethargy representa-tion. In this representation, equal areas below the curve correspondto equal neutron fluence rates in the considered energy region.

Clearly, there are only minor differences belowa neutron energyof about 1 MeV, while there are some major differences above: Theheight of the cascade peak at 100 MeV is much lower for theGEANT4 Bertini INC than that for GEANT4 Binary INC, MCNP/LAHET, and MCNP/HADRON due to the highest response valuesobtained by the GEANT4 Bertini INC in that energy range (Figs. 3and 4). In order to quantify these differences, the neutron fluencerates deduced from the four neutron spectra integrated over thefour energy regions indicated in Fig. 5 are given in Table 1.

As can be deduced from Table 1, for all codes/models the totalneutron fluence rates obtained do not differ much (less than 4%)from those obtained with the standard MCNP/LAHET responsefunctions. A closer look reveals, however, that the GEANT4ebasedresponse functions result in somewhat larger epithermal neutronfluence rates (12e13%) than those obtained with the MCNP-basedresponse functions. In contrast and more important in terms ofdose (see discussion below), the neutron fluence rate is signifi-cantly lower when based on GEANT4/Bertini INC (about 18%), forneutrons with energies above 20 MeV. This is because GEANT4/Bertini INC resulted in the highest response functions, at theseenergies (see Figs. 3 and 4). Altogether, the neutron spectra

10-2

10-1

100

101

10-9 10-7 10-5 10-3 10-1 101 103

mc[esnopse

R2 ]

Neutron Energy [MeV]

GEANT4 Binary INCGEANT4 Bertini INCMCNP/LAHET

9‘‘ Sphere (1‘‘ lead)

9‘‘ Sphere (0.5‘‘ lead)

Fig. 4. Response functions for those PE sphere that include lead, as obtained using thedifferent neutron transport codes and nuclear models.

0

20

40

60

80

100

120

10−9

10−7

10−5

10−3

10−1

101

103

d(E

mc[)

Ed/2−

h1−]


MCNP/LAHETMCNP/HADRONGEANT4 Binary INCGEANT4 Bertini INC

ThermalRegion

< 0.4 eV

CascadePeak

> 20 MeV

EpithermalRegion

0.4 eV - 0.1 MeV

EvaporationPeak

0.1 - 20 MeV

.

Fig. 5. Four neutron spectra unfolded from the mean count rates shown in Fig. 2, basedon the response functions calculated with GEANT4 Bertini INC, GEANT4 Binary INC,MCNP/HADRON and MCNP/LAHET, respectively.

Table 1Neutron fluence rates deduced from the neutron spectra of Fig. 5; percent valuesdescribe relative difference to the MCNP/LAHET results.

GEANT4/Bertini(cm!2 h!1)/(%)

GEANT4/Binary(cm!2 h!1)/(%)

MCNP/LAHET(cm!2 h!1)/(%)

MCNP/HADRON((cm!2 h!1)/(%)

Thermal 78.4/3.8 78.7/4.1 75.6/0 75.7/0.1Epithermal 80.4/13.4 79.7/12.4 70.9/0 70.7/!0.4Evaporation 69.1/!2.4 68.6/!3.1 70.8/0 71.1/0.4Cascade 107.4/!18.0 128.6/!1.8 131.0/0 127.2/!2.9Total 335.4/!3.7 355.7/2.1 348.3/0 344.6/!1.0

10-2

10-1

100

101

10-9 10-7 10-5 10-3 10-1 101 103

mc[esnopse

R2]


GEANT4 Binary INCGEANT4 Bertini INCMCNP/LAHETMCNP/HADRON

12‘‘ Sphere

8‘‘ Sphere

5‘‘ Sphere

Fig. 3. Response functions for the 5, 8, and 12 inch spheres, as obtained using thedifferent neutron transport codes and nuclear models.

C. Pioch et al. / Radiation Measurements 45 (2010) 1263e12671266

Pioch, C. et al. (2010). Rad Meas, 45(10), 1263–1267

erators were 0 cm ! (bare), 8.0 cm ! (1.5 cm thickness),11 cm ! (3.0 cm thickness), 15 cm ! (5.0 cm thickness),and 23 cm ! (9.0 cm thickness), respectively. The schematicdiagram of the measurement system is shown in Fig. 1. Allthe spherical 3He proportional counters with polyethylenemoderators were placed at 1m above the floor. High voltage(þ1;100V) was applied to each detector using high voltagepower supply (ORTEC model 478). Signals from each detec-tor were fed to the pre-amplifier (ORTEC model 142PC) andthe linear amplifier (ORTEC model 571). The output pulsesignals were introduced to a Multi Channel Analyzer(MCA) (ORTEC model 920E) to be converted into the pulseheight spectra. The pulse height spectra obtained by eachdetector were accumulated separately at every fixed time in-terval (1 h) by a personal computer, and summed up above

the cut-off level to discriminate the gamma-ray components.Then the total counts of each 3He counter were obtained bysumming up the accumulated counts above the cut-off levelover a given measurement time at each measuring location.

The response functions were required to convert themeasured counting rates to neutron energy spectrum. Theset of response functions for the energy range from thermalto 1GeV determined by Uwamino et al.8) were adoptedexcept for the bare 3He counter. The response functionof bare 3He counter was evaluated by Nunomiya et al.9)

Figure 2 shows the response functions used in the presentwork. The reliability of the response functions was con-firmed on an experimental basis. The calibration measure-ment of the BS was performed in the monoenergetic neutronfields with energy range of 250 keV to 15MeV at theDynamitron facility of Tohoku University,10) and also inthe thermal neutron field using a 252Cf neutron source andbare 252Cf neutron source in the Facility of RadiationStandards of JAERI (Japan Atomic Energy Research Insti-tute). Throughout the series of the calibration measurements,the differences between the measured and the calculatedvalues were within about 20%.

The cosmic-ray neutron energy spectra were obtainedby unfolding the measured counting rates with the aid ofresponse function. The SAND II code11) was used for thisunfolding, and the cosmic-ray neutron spectrum at sea levelobserved by Goldhagen et al.7) was employed as an initialguess spectrum for unfolding as shown in a solid histogramin Fig. 3. Table 1 shows the analytical conditions forSAND II in this study.

After obtaining neutron spectra by BS, the ambient doseequivalent rate H"(10) and the effective dose rate HE ofthe cosmic-ray at each altitude were estimated on the basisof ICRP Pubication74. In order to estimate the ambient doseequivalent rate H"(10), the neutron fluence to ambient dose

PreAmp.

PreAmp.

Pre Amp.

Pre Amp.

PreAmp.

H.V. Power Supply Linear

Amp.Linear Amp.

Linear Amp.

Linear Amp.

Linear Amp.

Multi Channel Analyzer P.C.

Fig. 1 Schematic diagram of the experimental arrangement of theBonner multi-sphere neutron spectrometer (BS)

0.001

0.01

0.1

1

10

100

1000

1.0E-02 1.0E+00 1.0E+02 1.0E+04 1.0E+06 1.0E+08

Neutron energy (eV)

Res

pons

e (c

ount

sne

utro

ns-1

cm-2

)

Calculated Measured23cm ϕ 23cm ϕ15cm ϕ 15cm ϕ11cm ϕ 11cm ϕ8cm ϕ 8cm ϕbare bare

Fig. 2 Measured and calculated response functions of the Bonner multi-sphere neutron spectrometer (BS)

496 M. KOWATARI et al.

JOURNAL OF NUCLEAR SCIENCE AND TECHNOLOGY

KOWATARI, M. et al. (2005). J. of Nuc. Sci. and Tech., 42(6), 495–502.

��10

LS Proton Recoil Detectors

CAEN Tools for Discovery

n

Application Note AN2506 Digital Gamma Neutron discrimination with Liquid Scintillators

1

Viareggio

19 November 2012

Introduction In recent years CAEN has developed a complete family of digitizers that consists of several models differing in sampling frequency, resolution, form factor and other features. Besides the use of the digitizers as waveform recorders (oscilloscope mode), CAEN offers the possibility to upload special versions of the FPGA firmware that implement algorithms for the Digital Pulse Processing (DPP); when the digitizer runs in DPP mode, it becomes a new instrument that represents a complete digital replacement of most traditional modules such as Multi Channel Analyzers, QDCs, TDCs, Discriminators and many others. In this application note, we describe the capability of the series x720 (12 bit, 250MSps) to perform neutron-gamma discrimination based upon the digital pulse shape analysis. The development of this FPGA firmware was based upon liquid scintillating detectors of type BC501-A. All detector and neutron/gamma source based tests were performed at the Triangle Universities Nuclear Laboratory (TUNL) on the campus of the Duke University in collaboration with Mohammad Ahmed.

Neutron-Gamma discrimination with liquid scintillators Liquid scintillating detectors are widely used to achieve neutron-gamma discrimination due to their effective Pulse Shape Discrimination (PSD) properties [1]. The light emission of liquid scintillating detectors comprises of a fast decay component, as well as a substantial slow decay component. These components arise from de-excitations of different atomic states in the scintillator. The relative population of these states depends on the energy loss (dE/dx) of the particle. In organic liquid scintillators, these components strongly depend on the energy loss. The gamma rays interact in the scintillator mostly via atomic Compton scattering or pair production mechanisms. The neutrons are detected by scattering them with the protons. These two different processes for gamma-rays and neutrons give rise to significant difference in the slow decay component of the light emission. This difference becomes the basis of pulse shape discrimination in the liquid scintillating detectors [2].

Set-up description

The detector is a 5 inch diameter, 2 inch thick liquid scintillator of type BC501-A. The scintillator is coupled to a Hamamatsu R1250 photo-multiplier tube (PMT). The base is also a Hamamatsu model HTV H6527. The detector is biased to -1430V. The signals from the PMTs are connected to the digitizer inputs via 120 feet low-loss RG8 cables. The detector is biased to -1430V. The signals from the PMTs are connected to the digitizer inputs via 120 feet low-loss RG8 cables. The digitizer is a 4 channel, 12 bit, 250 MS/s DT5720 (Desktop version). The analog bandwidth is about 120 MHz and the input dynamic range is 2 Vpp; the DC offset can be adjusted by means of an internal DAC in the range -2/+2 V. The digitizer runs the preliminary version of the DPP-PSD firmware for the digital charge integration and neutron-gamma discrimination. This principle of operation of the DPP-PSD firmware can be summarized by the following operations that are executed in real time inside the FPGA

x The baseline of the signal is calculated by a programmable length mean filter and subtracted from the input signal; the result is compared with the trigger threshold and, if exceeded, a trigger is issued; it is also possible to use the threshold only to get armed, then wait for the pulse peak and issue the trigger at that time.

x With the trigger, the baseline is frozen, two gates of programmable width are opened and the charge integration starts; the signal fed to the integrator is delayed of a programmable number of samples in order to let the gates to start before the trigger.

x The trigger also initiates the event building: this can include the waveforms (i.e. row samples) of the input, baseline, trigger, gate and other signals, the time stamp of the trigger and the charges calculated within the two gates (short and long).

Fig. 1: The detector used for these measures is a 5”x2” BC501-A liquid scintillator coupled to an Hamamatsu R1250 PMT

Liquid Scintillator

2.5-MeV source. We determined the templates by minimizing thesquared Euclidean distance (L2) of the normalized pulses withineach of the two clusters. We also estimated templates with arobust version of cluster analysis based on an L1 distance metric.In this approach, within each cluster, the median value ratherthan the mean value is computed. The robust and non-robustcluster analysis methods yield similar template estimates.

From the 137Cs gamma-ray source, we determined an electronicrecoil template by a robust signal averaging method. Each baseline-corrected pulse was normalized so that its maximum value was 1.At each time sample, the trimmed mean of all the processed pulseswas computed, and the resulting pulse was divided by its integralvalue. Values of the trimmed mean at each relative time of interestbetween the 0.1 and 0.9 quantiles of the distribution were averaged.For the 2.5-MeV source, we estimated a nuclear recoil template withthe same robust signal averaging method described above. Theestimated nuclear recoil and electronic recoil templates from thecluster analysis agree well with the corresponding robust signalaveraging estimates. Moreover, the estimated nuclear recoil tem-plates determined from start and stop pulses for the 2.5 MeV casewere in very close agreement for the range of amplitudes that weattribute to neutron capture on 6Li.

4.3. Discrimination statistics

The Matusita distances between a normalized pulse of interest, pm,and the template pulses for the electronic recoil pe and nuclear recoilevents pn are

de ¼X

i

ffiffiffiffiffiffiffiffiffiffiffipmðiÞ

p$

ffiffiffiffiffiffiffiffiffiffipeðiÞ

q" #2

ð6Þ

and

dn ¼X

i

ffiffiffiffiffiffiffiffiffiffiffipmðiÞ

p$

ffiffiffiffiffiffiffiffiffiffipnðiÞ

q" #2

ð7Þ

where i is the time increment for the digitized pulse. The normalizedpulses sum to 1. Negative values are set to 0 before taking squareroots in the above equations. Our primary PSD statistic is

logR¼ logdn

de: ð8Þ

For comparison, we also computed a prompt ratio statistic

fp ¼Xp

XTð9Þ

where Xp is the integrated pulse from t¼0 to to and XT is theintegrated pulse over all times. Here, we set to to be the timewhere the nuclear and electronic recoil pulses cross.

For both discrimination statistics, Figs. 8 and 9, we estimate anamplitude dependent discrimination threshold based on events thatproduce logR values less than 0. We then formed a curve in(amplitude, logR) or (amplitude, fp) space. For each method, wesorted the corresponding curve data according to amplitude bins anddetermine the median amplitude and median discrimination statisticwithin each bin. In sequence, we fit a monotonic regression model[52] and then a smoothing spline to each curve. The degrees offreedom of the smoothing spline were determined by cross-valida-tion [53]. We determined a threshold for each particular amplitudeby evaluating the smoothing spline model at that amplitude.

The separation between the logR statistics appears more dramaticthan the separation between the fp statistics for the 137Cs and2.5-MeV sources. Theoretically, we expect that the logR statisticconveys more information because it is based on a 201-bin repre-sentation of the observed pulse whereas the prompt ratio is based ona 2-bin representation of the observed pulse. A careful quantificationof the relative performance of PSD algorithms based on these twostatistics is a topic for further study. One could also form larger binsto smooth out noise before computing a logR statistic for any pulse asdiscussed in Refs. [54,55]. In future experiments, our digital acquisi-tion system will have a higher (10-bit or 12-bit) resolution comparedto the 8-bit resolution of the data shown in this study. This shouldfacilitate refinement of our PSD techniques. In this work, weneglected to account for the energy dependence of the templates.In future work, we may account for this dependence.

5. Summary and conclusions

A liquid scintillator doped with 0.15% 6Li by weight was fabricatedand made into a test cell. The process of making the scintillator does

Fig. 7. Waveform templates for nuclear recoil and electronic recoil eventsdetermined by cluster analysis from calibration data from a 2.5-MeV neutronsource (contaminated by gamma-rays).

Fig. 8. Empirical distribution of logR statistics.

Fig. 9. Empirical distribution of prompt ratio statistics. The width of the prompt timewindow is determined by where the nuclear and electronic recoil templates cross.

B.M. Fisher et al. / Nuclear Instruments and Methods in Physics Research A 646 (2011) 126–134 133

PSD

• N-P scattering to detect deposited energy

• Use pulse shape to separate gammas from neutrons

• All neutron interactions detected, mostly partial-energy depositions


131210 1198765432

1MeV

Light Output (MeVee)

Resp

onse

Fun

ctio

n (A

rb. U

nit)

14MeV

1 2 3 4 5 6 7

1

0.01

0.1

1.3.2 The liquid scintillator proton recoil detector

The liquid scintillator proton recoil detector is perhaps the opposite of the

passively moderated 3He counter. Rather than detecting the capture of a thermal-

ized neutron, these detectors function by detecting the recoil proton from a neutron

scatter. With their large hydrogen content, liquid organic scintillators are highly

e↵ective neutron moderators. Depending upon the mass of the recoiling nucleus, a

neutron can transfer any fraction of it’s energy, up to a certain cuto↵ energy.

Er =4A

(1 + A)2(cos2 ✓)En (1.8)

Er|max =4A

(1 + A)2En (1.9)

Er is the energy of the recoiling nucleus, A is the atomic number of the nucleus, ✓

is the angle (in the lab frame) of the recoil, and En is the incident neutron energy.

For hydrogen, the neutron can transfer all of it’s energy to the recoiling proton. For

carbon, the other main element in organic scintillator, the recoil nucleus can receive

a maximum of 0.28 En [12]. Carbon recoils do not produce light in the scintillator,

but manifest as a loss of up to 28% of the total neutron energy.

Liquid scintillator also possesses the important ability to distinguish between

nuclear recoils and electronic recoils using the shape of the resulting pulse of light.

Nuclear recoils excite longer-lived excitations (triplet vs singlet state) in liquid scin-

tillator which lead to a slightly longer pulse of light. Figure 1.19 shows example

traces of typical gamma and neutron interactions in liquid organic scintillator. The

29

��11

Fisher et al. 2011

Ido et al. 2009

C

AE

N

Tool

s fo

r D

isco

very

n

Appl

icat

ion

Not

e AN

2506

Di

gita

l Gam

ma

Neu

tron

disc

rimin

atio

n w

ith Li

quid

Scin

tilla

tors

1

Vi

areg

gio

19 N

ovem

ber 2

012

Intr

oduc

tion

In re

cent

yea

rs C

AEN

has

dev

elop

ed a

com

plet

e fa

mily

of d

igiti

zers

that

con

sists

of s

ever

al m

odel

s di

fferin

g in

sam

plin

g fr

eque

ncy,

re

solu

tion,

for

m f

acto

r an

d ot

her

feat

ures

. Bes

ides

the

use

of

the

digi

tizer

s as

wav

efor

m r

ecor

ders

(os

cillo

scop

e m

ode)

, CAE

N

offe

rs t

he p

ossib

ility

to

uplo

ad s

peci

al v

ersio

ns o

f the

FPG

A fir

mw

are

that

impl

emen

t al

gorit

hms

for

the

Digi

tal P

ulse

Pro

cess

ing

(DPP

); w

hen

the

digi

tizer

run

s in

DPP

mod

e, it

bec

omes

a n

ew in

stru

men

t tha

t rep

rese

nts

a co

mpl

ete

digi

tal r

epla

cem

ent o

f mos

t tr

aditi

onal

mod

ules

suc

h as

Mul

ti Ch

anne

l Ana

lyze

rs, Q

DCs,

TDC

s, D

iscrim

inat

ors

and

man

y ot

hers

. In

this

appl

icat

ion

note

, we

desc

ribe

the

capa

bilit

y of

the

serie

s x7

20 (1

2 bi

t, 25

0MSp

s) to

per

form

neu

tron

-gam

ma

disc

rimin

atio

n ba

sed

upon

the

digi

tal p

ulse

sh

ape

anal

ysis.

The

dev

elop

men

t of t

his F

PGA

firm

war

e w

as b

ased

upo

n liq

uid

scin

tilla

ting

dete

ctor

s of t

ype

BC50

1-A.

Al

l det

ecto

r and

neu

tron

/gam

ma

sour

ce b

ased

test

s w

ere

perf

orm

ed a

t the

Tria

ngle

Uni

vers

ities

Nuc

lear

Lab

orat

ory

(TU

NL)

on

the

cam

pus o

f the

Duk

e U

nive

rsity

in c

olla

bora

tion

with

Moh

amm

ad A

hmed

.

Neu

tron

-Gam

ma

disc

rimin

atio

n w

ith li

quid

scin

tilla

tors

Li

quid

sci

ntill

atin

g de

tect

ors

are

wid

ely

used

to

achi

eve

neut

ron-

gam

ma

disc

rimin

atio

n du

e to

the

ir ef

fect

ive

Pulse

Sha

pe

Disc

rimin

atio

n (P

SD) p

rope

rtie

s [1]

. Th

e lig

ht e

miss

ion

of l

iqui

d sc

intil

latin

g de

tect

ors

com

prise

s of

a f

ast

deca

y co

mpo

nent

, as

wel

l as

a s

ubst

antia

l slo

w d

ecay

co

mpo

nent

. The

se c

ompo

nent

s ar

ise f

rom

de-

exci

tatio

ns o

f di

ffere

nt a

tom

ic s

tate

s in

the

sci

ntill

ator

. The

rel

ativ

e po

pula

tion

of

thes

e st

ates

dep

ends

on

the

ener

gy lo

ss (d

E/dx

) of t

he p

artic

le. I

n or

gani

c liq

uid

scin

tilla

tors

, the

se c

ompo

nent

s str

ongl

y de

pend

on

the

ener

gy lo

ss.

The

gam

ma

rays

inte

ract

in th

e sc

intil

lato

r mos

tly v

ia a

tom

ic C

ompt

on s

catt

erin

g or

pai

r pro

duct

ion

mec

hani

sms.

The

neu

tron

s ar

e de

tect

ed b

y sc

atte

ring

them

with

the

prot

ons.

The

se tw

o di

ffere

nt p

roce

sses

for

gam

ma-

rays

and

neu

tron

s gi

ve r

ise to

sig

nific

ant

diffe

renc

e in

the

slo

w d

ecay

com

pone

nt o

f the

ligh

t em

issio

n. T

his

diffe

renc

e be

com

es t

he b

asis

of p

ulse

sha

pe d

iscrim

inat

ion

in

the

liqui

d sc

intil

latin

g de

tect

ors [

2].

Set-

up d

escr

iptio

n

The

dete

ctor

is a

5 in

ch d

iam

eter

, 2 in

ch th

ick

liqui

d sc

intil

lato

r of t

ype

BC50

1-A.

The

sci

ntill

ator

is

coup

led

to a

Ham

amat

su R

1250

pho

to-

mul

tiplie

r tu

be (

PMT)

. Th

e ba

se i

s al

so a

Ham

amat

su m

odel

HTV

H6

527.

The

det

ecto

r is b

iase

d to

-143

0V.

The

signa

ls fr

om t

he P

MTs

are

con

nect

ed t

o th

e di

gitiz

er in

puts

via

12

0 fe

et lo

w-lo

ss R

G8 c

able

s.

The

dete

ctor

is

bias

ed t

o -1

430V

. Th

e sig

nals

from

the

PM

Ts a

re

conn

ecte

d to

the

digi

tizer

inpu

ts v

ia 1

20 fe

et lo

w-lo

ss R

G8 c

able

s.

The

digi

tizer

is

a 4

chan

nel,

12 b

it, 2

50 M

S/s

DT57

20 (

Desk

top

vers

ion)

. Th

e an

alog

ban

dwid

th i

s ab

out

120

MHz

and

the

inp

ut

dyna

mic

rang

e is

2 Vp

p; th

e DC

offs

et c

an b

e ad

just

ed b

y m

eans

of a

n in

tern

al D

AC in

the

rang

e -2

/+2

V.

The

digi

tizer

runs

the

prel

imin

ary

vers

ion

of th

e DP

P-PS

D fir

mw

are

for

the

digi

tal c

harg

e in

tegr

atio

n an

d ne

utro

n-ga

mm

a di

scrim

inat

ion.

Thi

s pr

inci

ple

of o

pera

tion

of th

e DP

P-PS

D fir

mw

are

can

be su

mm

arize

d by

th

e fo

llow

ing

oper

atio

ns t

hat

are

exec

uted

in

real

tim

e in

side

the

FPGA

x Th

e ba

selin

e of

the

sig

nal i

s ca

lcul

ated

by

a pr

ogra

mm

able

leng

th m

ean

filte

r an

d su

btra

cted

from

the

inpu

t sig

nal;

the

resu

lt is

com

pare

d w

ith th

e tr

igge

r thr

esho

ld a

nd, i

f exc

eede

d, a

trig

ger i

s iss

ued;

it is

also

pos

sible

to u

se th

e th

resh

old

only

to g

et a

rmed

, the

n w

ait f

or th

e pu

lse p

eak

and

issue

the

trig

ger a

t tha

t tim

e.

x W

ith th

e tr

igge

r, th

e ba

selin

e is

froz

en, t

wo

gate

s of

pro

gram

mab

le w

idth

are

ope

ned

and

the

char

ge in

tegr

atio

n st

arts

; th

e sig

nal f

ed to

the

inte

grat

or is

del

ayed

of a

pro

gram

mab

le n

umbe

r of s

ampl

es in

ord

er to

let t

he g

ates

to s

tart

bef

ore

the

trig

ger.

x Th

e tr

igge

r al

so in

itiat

es t

he e

vent

bui

ldin

g: t

his

can

incl

ude

the

wav

efor

ms

(i.e.

row

sam

ples

) of

the

inpu

t, ba

selin

e,

trig

ger,

gate

and

oth

er s

igna

ls, th

e tim

e st

amp

of th

e tr

igge

r and

the

char

ges

calc

ulat

ed w

ithin

the

two

gate

s (s

hort

and

lo

ng).

Fig.

1: T

he detec

tor u

sed for t

hese m

easu

res is a

5”x2”

BC5

01-A

liq

uid

scin

tilla

tor c

oupl

ed to

an

Ham

amat

su R

1250

PM

T

Bonner Spheres LS Proton Recoil

• Only measure thermal neutron capture

• Very difficult to calibrate response functions

• No direct energy observation

• Detect energy from every interaction

• PID from pulse shape • Most events are partial energy

deposition, bad for spectroscopy

Vs


erators were 0 cm ! (bare), 8.0 cm ! (1.5 cm thickness),11 cm ! (3.0 cm thickness), 15 cm ! (5.0 cm thickness),and 23 cm ! (9.0 cm thickness), respectively. The schematicdiagram of the measurement system is shown in Fig. 1. Allthe spherical 3He proportional counters with polyethylenemoderators were placed at 1m above the floor. High voltage(þ1;100V) was applied to each detector using high voltagepower supply (ORTEC model 478). Signals from each detec-tor were fed to the pre-amplifier (ORTEC model 142PC) andthe linear amplifier (ORTEC model 571). The output pulsesignals were introduced to a Multi Channel Analyzer(MCA) (ORTEC model 920E) to be converted into the pulseheight spectra. The pulse height spectra obtained by eachdetector were accumulated separately at every fixed time in-terval (1 h) by a personal computer, and summed up above

the cut-off level to discriminate the gamma-ray components.Then the total counts of each 3He counter were obtained bysumming up the accumulated counts above the cut-off levelover a given measurement time at each measuring location.

The response functions were required to convert themeasured counting rates to neutron energy spectrum. Theset of response functions for the energy range from thermalto 1GeV determined by Uwamino et al.8) were adoptedexcept for the bare 3He counter. The response functionof bare 3He counter was evaluated by Nunomiya et al.9)

Figure 2 shows the response functions used in the presentwork. The reliability of the response functions was con-firmed on an experimental basis. The calibration measure-ment of the BS was performed in the monoenergetic neutronfields with energy range of 250 keV to 15MeV at theDynamitron facility of Tohoku University,10) and also inthe thermal neutron field using a 252Cf neutron source andbare 252Cf neutron source in the Facility of RadiationStandards of JAERI (Japan Atomic Energy Research Insti-tute). Throughout the series of the calibration measurements,the differences between the measured and the calculatedvalues were within about 20%.

The cosmic-ray neutron energy spectra were obtainedby unfolding the measured counting rates with the aid ofresponse function. The SAND II code11) was used for thisunfolding, and the cosmic-ray neutron spectrum at sea levelobserved by Goldhagen et al.7) was employed as an initialguess spectrum for unfolding as shown in a solid histogramin Fig. 3. Table 1 shows the analytical conditions forSAND II in this study.

After obtaining neutron spectra by BS, the ambient doseequivalent rate H"(10) and the effective dose rate HE ofthe cosmic-ray at each altitude were estimated on the basisof ICRP Pubication74. In order to estimate the ambient doseequivalent rate H"(10), the neutron fluence to ambient dose

PreAmp.

PreAmp.

Pre Amp.

Pre Amp.

PreAmp.

H.V. Power Supply Linear

Amp.Linear Amp.

Linear Amp.

Linear Amp.

Linear Amp.

Multi Channel Analyzer P.C.

Fig. 1 Schematic diagram of the experimental arrangement of theBonner multi-sphere neutron spectrometer (BS)

0.001

0.01

0.1

1

10

100

1000

1.0E-02 1.0E+00 1.0E+02 1.0E+04 1.0E+06 1.0E+08

Neutron energy (eV)

Res

pons

e (c

ount

sne

utro

ns-1

cm-2

)

Calculated Measured23cm ϕ 23cm ϕ15cm ϕ 15cm ϕ11cm ϕ 11cm ϕ8cm ϕ 8cm ϕbare bare

Fig. 2 Measured and calculated response functions of the Bonner multi-sphere neutron spectrometer (BS)

496 M. KOWATARI et al.

JOURNAL OF NUCLEAR SCIENCE AND TECHNOLOGY

��12

PMT PM

T

Capture Gated Spectroscopy

pt

p

500

400

300

200

Chan

nel

155154153152151150149148x10

3

pp

4000

3950

3900

3850

Chan

nel

36.98x10336.9636.9436.9236.9036.8836.86Samples

4000

3950

3900

3850

Chan

nel

36.98x10336.9636.9436.9236.9036.8836.86Samples

4000

3950

3900

3850

Chan

nel

36.98x10336.9636.9436.9236.9036.8836.86Samples

Combine features from BS and PR detectors: 1.Demand capture

- Full energy deposition 2.Directly measure deposited energy


n Scintillator

3He

��13

The FaNS Detectors


The FaNS Detectors• Arrays of plastic scintillator segments and 3He

proportional counters

• Segmentation improved energy reconstruction

• Use Capture-gated Spectroscopy for particle identification and energy information

• Calibrated at NIST with Cf-252, DD, and DT neutrons

• FaNS-1: Rapidly deployed as proof-of-principle

• FaNS-2: Full fledged detector


30

25

20

15

10

5

0Li

ght

Units

(M

eVee

)302520151050

Energy (MeV)

Linear Response Proton Response

Energy Resolution Through Segmentation

• Plastic scintillator has a non-linear light response to heavy charged particles

• To achieve better energy resolution, we segment the detector to reconstruct each scatter separately

• Segment size is chosen to match the mean free path of target neutrons

We want to detect each scatter separately, convert light into energy, and then sum

WIDG Seminar 2/18/2014T.J. Langford ��16

CGS with FaNS

Proton Recoils

3He Neutron Capture�t

4000

3000

2000

1000

0

-1000

Chan

nel (

arb)

-15 -10 -5 0 5 10 15Time (us)

Block8, PMT1 Block8, PMT2 Block12, PMT1 Block12, PMT2 He3

• Trigger on the 3He capture • Record scintillator signals that

occur within a fixed time before the capture

• Negative coincidences must be random


Time Separation Spectrum


Random Coincidences

Real Coincidences

Time Separation Spectrum


FaNS-1 Summary

• 18liters of PS in six segments • Six He-3 proportional counters • Calibrated at NIST • Measured surface spectrum to 150MeV • Installed at Kimballton Lab • Measured the neutron spectrum and flux at KURF

700

600

500

400

300

200

100

0

Coun

ts (

arb)

543210Energy (MeV)

Data Monte Carlo

600

500

400

300

200

100

0

Coun

ts (

arb)

151050Pulse Ht (MeV)

Data Monte Carlo

��20T.J. Langford WIDG Seminar 2/18/2014

DD

DT

FaNS-1Surface Neutrons

10-7

10-6

10-5

10-4

10-3

10-2

10-1

Coun

ts /

MeV

/s

5 6 7 8 91

2 3 4 5 6 7 8 910

2 3 4 5 6 7 8 9100

2 3

Energy (MeV)

BG Subtracted Scaled MCNP


�(En > 1 MeV ) = (4.0± 1.0)⇥ 10�3n/cm2/s

��21

FaNS-1 at KURF - 1450 mwe

3

4567

1

2

3

4567

10

2

3

4567

100

Coun

ts/M

eV

8642Neutron Energy (MeV)

Real Coincidences Random Coincidences

1.0

0.8

0.6

0.4

0.2

0.0

Rise

Tim

e (u

s)

543210Pulse Ht (MeV)

↵

µD

n

�

• 3.74x106 s dataset • 1.67x105 He-3 triggers • 384 counts pass all cuts • 89 remain after BG

subtraction

WIDG Seminar 2/18/2014T.J. Langford

30

20

10

0

Coun

ts

-200 -100 0 100 200Time Separation (µs)

��22

FaNS-1 at KURF - 1450mwe

8910-7

2

3

4

5678910-6

2

3

4

5678910-5

Coun

ts/M

eV/s

8642Neutron Energy (MeV)

BG Subtracted


15

events/keV-kg-year, is precisely what we have found inour simulation.

Depth (km.w.e.)1 2 3 4 5 6

)-1 y

ear

-1 k

g-1

Even

ts (k

eV

-710

-610

-510

-410

-310

-210

-110

12510

2610

2710

2810

Hal

f Life

(yea

rs)

Majorana

Granularity, PSD and Segmentation

Target sensitivity

Without neutron veto

With neutron veto (99%)

WIP

P

Gra

n Sa

sso

Sudb

ury

FIG. 25: The Depth-Sensitivity-Relation (DSR) derived fora Majorana-like experiment showing, specifically, the resultsfrom this work assuming the detector is operated at a depthequivalent to the Gran Sasso Laboratory. The raw event ratein the energy region of interest of 0.026 events/keV-kg-yearcan be reduced by a factor of 7.4 by exploiting the detectorgranularity, pulse-shape discrimination (PSD), and detectorsegmentation. The upper curve displays the background sim-ulated in the case that no active neutron veto is present andthe lower curve indicates the reduction that would ensue if anactive neutron veto were present that is 99% efficient.

The results of our simulations can be used to derivethe DSR for Majorana as shown in Fig. 25. The neu-tron induced background can be reduced by about a fac-tor of 7.4 in Majorana owing to the use of crystal-to-crystal coincidences and the use of pulse-shape discrim-ination and segmentation. Nonetheless, to achieve thetarget sensitivity of next generation double-beta decayexperiments, 0.00025 events/keV-kg-year correspondingto the background level required to reach sensitivity tothe atmospheric mass scale of 45 meV Majorana neutrinomass, the muon-induced background must be reduced byroughly another factor of 100. This can be achieved onlyby operating such a detector at depths in excess of 5km.w.e., otherwise an active neutron veto would need tobe implemented with an efficiency in excess of 99%.

D. (α,n) Background

Once the depth requirement is satisfied, a proper shieldagainst (α,n) neutrons from the environment becomesnecessary. We use the standard rock and the measuredneutron flux (3.78×10−6cm−2s−1 [68, 69] ) at Gran Sassoassuming that all underground labs have the same orderof neutron flux to establish the shielding requirement for(α, n) neutrons. This flux corresponds to an average ofabout 2.63 ppm 238U and 0.74 ppm 232Th activity inGran Sasso rock and 1.05 ppm 238U and 0.67 ppm 232Thactivity in Gran Sasso concrete [70]. The neutron energy

spectrum depending on the rock composition is shown inFig. 26. As can be seen, the total neutron flux is aboutthree orders of magnitude higher than that of neutronsfrom the rock due to muon-induced processes, but theenergy spectrum is much softer.

Neutron Energy (MeV)0 1 2 3 4 5 6 7 8 9 10

)-1 M

eV-1

s-2

Neu

tron

s (c

m

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

0.2

0.22-610×

Standard Rock

Gran Sasso Rock

FIG. 26: The neutron energy spectrum arising from (α,n)reactions due to radioactivity in the rock. We predict a harderenergy spectrum in Gran Sasso rock relative to standard rockowing to the presence of carbon and magnesium.

To demonstrate the neutron flux and energy spectrumat different boundaries we show the rock/cavern neutronflux and energy spectrum with a shielding for Majoranadescribed earlier in Fig. 27 for the depth of Gran Sassoas an example.

Energy (MeV)-110 1 10 210 310

)-1 M

eV-1

y-2

Neu

tron

s (c

m

-1210-1110

-1010

-910

-810

-710

-610

-510

-410

-310

-210

-1101

10

210






FIG. 27: The energy spectrum for fast neutrons producedby (α, n) reactions in the rock compared to those induced bymuon interactions in the rock with and without shielding. Thelower energy neutrons (< 10 MeV) are quickly absorbed usingpolyethylene shielding, however, the high energy portion ofthe muon-induced neutron flux persists. The addition of leadshielding adjacent to a detector can also create an additionalsource of muon-induced neutrons.

Note that the (α, n) neutrons from the rock are quicklyattenuated to the level of the muon-induced neutrons be-

(↵, n)

��23

�(En > 1.4 MeV) = (1.6± 0.4)⇥ 10�6 n/cm2/s

FaNS-1 Conclusions• Demonstrates the power of CGS

• Detect peaks rather than edges

• Measured the surface fast neutron spectrum and flux from 1 to 150 MeV

• Deployed FaNS-1 at KURF, where an (alpha, n)-like spectrum and flux were successfully measured

• Influenced design of FaNS-2

• Focus on muon-induced neutrons (En > 10 MeV)

• Operate at a shallow location to measure flux and spectrum


FaNS-2



FaNS-2

��26


FaNS-221 3He Neutron

Detectors

9cm X 9cm X 56cm Plastic Scintillator Bars

��27

FaNS-2• The “Purpose-built” Detector

• Improvement over FaNS-1 in almost every way:

• Volume: 15 liters → 72 liters

• DAQ: twice as fast, 8ch → 56ch

• PMTs: better linearity and characterization

• Built at UMD, commissioned at NIST in December 2012

• FaNS-2 was operated in a low scatter room measuring source and ambient neutrons

• Reduce backscattering neutrons compared to FaNS-1

• Effectively no shielding of ambient neutrons from cosmic rays


Cf-252 Source Calibration2.5

2.0

1.5

1.0

0.5

0.0

Coun

t ra

te (

n/s)

> 2

MeV

25x10-320151050Solid Angle Subtended

MCNP Corrected Data

Dataa =0.04 ± 0.03b =77.0 ± 2.5

MCNPa =0.035 ± 0.007b =80.6 ± 0.8

• Seven different source positions were measured - (85 cm - 238 cm above FaNS-2) and two runs without the source

• Each position was modeled in MCNP, and the same cuts were applied to data and simulation

• The slope of the fit-line is extracted for the efficiency


Cf

✏ =slope

source activity

= (3.6± 0.15)%

��29


Target AcceleratingPotential

Drift RegionHV

Feedthrough

Neutron Generator Data

D +D ! 3He+ n (2.5 MeV )D + T ! 4He+ n (14.1 MeV )

2500

2000

1500

1000

500

0

Coun

ts

43210Energy (MeV)

Data MCNPDD

600

400

200

0

Coun

ts

20151050Energy (MeV)

Data MCNPDT

��30

Cosmogenic Neutrons at NISTn

nn

Isotropic neutron field

Gordon et al. Surface Neutron Spectrum

• Determine the response of FaNS-2 to an assumed cosmogenic neutron spectrum • Use the response to convert measured neutrons into incident flux

✏ = 87± 13 n/(fluence)


• Operated periodically over the course of 6 months

• 4x106 triggers were collected

• Post BG subtraction,1.7x106 events remain

Cosmogenic Neutrons at NIST


10-8

10-7

10-6

10-5

10-4

10-3

10-2

10-1

Coun

ts /

MeV

/s

100 101 102 103 104

Energy (MeV)

Real+Random Random Only

2.5x10-3

2.0

1.5

1.0

0.5

0.0

Coun

ts/s

6004002000-200Time Separation (µs)

��32

1050

1040

1030

1020

1010

1000

Barometric Pressure (m

b)

3/16/13 3/21/13 3/26/13 3/31/13 4/5/13 4/10/13Date

0.70

0.65

0.60

0.55

0.50

0.45

0.40

FaNS

-2 C

ount

Rat

e (n

/s)

Barometric Variation

��33

0.70

0.68

0.66

0.64

0.62

0.60

0.58

0.56

0.54

FaNS

-2 B

GSub

Rat

e (1

/s)

3/16/13 3/21/13 3/26/13 3/31/13 4/5/13 4/10/13Date

105

100

95

90

85

Newark Neutron M

onitor Rate (1/s)

NMDB FaNS-2

100

90

80

70

60

Neu

tron

Mon

itor R

ate

(n/s

)

3/16/13 3/21/13 3/26/13 3/31/13 4/5/13 4/10/13Date

1050

1040

1030

1020

1010

1000

Barometric Pressure (m

b)

http://www.nmdb.eu/T.J. Langford WIDG Seminar 2/18/2014

• Neutron monitoring stations to study solar activity are spread across the globe

• Data freely available!

• Can compare pressure corrected and uncorrected event rates

94

92

90

88

86

Newark Neutron M

onitor Rate (1/s)

3/16/13 3/21/13 3/26/13 3/31/13 4/5/13 4/10/13Date

0.66

0.64

0.62

0.60

0.58

0.56

FaNS

-2 B

GSub

Rat

e (1

/s)

NMDB FaNS-2

http://www.nmdb.eu/


10-8

10-7

10-6

10-5

10-4

10-3

10-2

10-1

Coun

ts /

MeV

/s

100 101 102 103 104

Energy (MeV)

Data MCNP

�(n) =(1.753± 0.003)⇥ 106 n

4.842⇥ 106 s⇥ (87± 13) n/(n/cm2)

�(n) = (4.16± 0.6)⇥ 10�3 n/cm2/s

Flux above 2 MeV


��34



2.5x10-3

2.0

1.5

1.0

0.5

0.0

E x

dφ/d

E (n

/s)

100 101 102 103 104

Energy (MeV)

Data MCNP

Directly observe features at 1.5 and 4 MeV that

BBS cannot resolve

��35

CGS Measurement at LNGS

Direct measurement of the atmospheric neutron flux in the energyrange 10–500 MeV

Antonio Bonardi a,b,c,*, Marco Aglietta b,c, Gianmarco Bruno d,e, Walter Fulgione b,c,Ana Amelia Bergamini Machado e

a Università di Torino: via Giuria 1, 10125 Torino, Italyb INFN Torino: via Giuria 1, 10125 Torino, Italyc IFSI Torino: Corso Fiume 4, 10133 Torino, Italyd Università de l’Aquila: Via Vetoio Località Coppito, 67100 L’Aquila, Italye INFN-Laboratori Gran Sasso: S.S. 17 BIS km 18.910, 67010 Assergi L’Aquila, Italy

a r t i c l e i n f o

Article history:Received 24 May 2010Received in revised form 27 July 2010Accepted 28 July 2010Available online 5 August 2010

Keywords:NeutronsFluxMeasurement

a b s t r a c t

The results of a direct measurement of the atmospheric neutron flux in the energy range 10–500 MeVperformed at 42!2501100 N, 13!310200 E, rigidity cutoff 6.3 GV, altitude 970 m a.s.l. (LNGS external site)on November 2008, during minimum solar activity, are reported.

The detector consists of a 1.5 ! 1 ! 1 m3 stainless steel tank filled with 1.2 tons of 0.1% Gd doped liquidscintillator, monitored by three photomultipliers and surrounded by a 4p active muon veto. The mea-surement is performed by observing events formed by two un-vetoed pulses inside a temporal window95 ls long: the first one due to a recoiling proton scattered by a neutron, the second one due to the neu-tron capture (n,Gd).

The resulting atmospheric neutron fluxes are:

UðE > 10 MeVÞ ¼ 47% 5 neutrons s&1 m&2;

UðE > 20 MeVÞ ¼ 42% 4 neutrons s&1 m&2:

" 2010 Elsevier B.V. All rights reserved.

1. Introduction

All the neutrons produced by the interactions of galactic and so-lar cosmic rays in the atmosphere are called ‘‘atmospheric neu-trons”. They are generated in hadronic and electromagnetic airshowers by spallation and evaporation processes on Nitrogen andOxygen nuclei.

Neutrons can be dangerous for electronic devices and humanhealth at aviation altitude where they contribute to about half ofthe dose equivalent of the secondary cosmic radiation [1–3],whereas at sea level they are a significant background source forseveral types of surface detectors. In particular our attention hereis focused on studying the atmospheric neutron background for aliquid scintillator antineutrino surface detector.

Since the famous Cowan and Reines experiment [4], nuclearpower plants have been used for studying neutrino properties

and nowadays neutrino oscillation experimental observations arereactivating this experimental approach.

In principle, it is possible to turn the tide, i.e. to use antineutrinodetection for monitoring the activity of nuclear power plants:this may be very useful in particular for detecting illicit or suspi-cious uses of these facilities [5]. The International Atomic EnergyAgency is in the process of developing new technologies to monitornuclear activity on power plants, aiming for non-proliferationsafeguards.

The advantages of using !me for reactor safeguards and monitoringinclude the availability of real time information on the status of thecore, less intrusiveness, and simplified operations from the stand-point of both the reactor operator and the safeguards agency [5].

Such a !me detector has to be small ('1 m3) [5] and not overbur-dened, due to cost and logistical reasons: that means a very poorshielding factor for cosmic radiation.

Up to now, !me detectors are designed to look for Inverse Beta De-cay (IBD) !me þ p) eþ þ n and their active mass is monitored by pho-tomultiplier tubes (PMTs). In case of IBD the prompt e+ signal isfollowed by a delayed one due to neutron capture: the double signa-ture allows the disentangling of !me signals from random background.

0927-6505/$ - see front matter " 2010 Elsevier B.V. All rights reserved.doi:10.1016/j.astropartphys.2010.07.004

Corresponding author at: Università di Torino, Via Giuria 1, 10125 Torino, Italy.Tel.: +39 3497942509.

E-mail address: [email protected] (A. Bonardi).

Astroparticle Physics 34 (2010) 225–229

Contents lists available at ScienceDirect

Astroparticle Physics

journal homepage: www.elsevier .com/ locate/ast ropart

(MeV)visE10 210

)−

1flu

x (H

z M

eV

−310

−210

−110

1

• 1.2 m3 liquid scintillator doped with Gd, operated for 5 days

• Collected ~6x103 neutron events

• Measured flux above two thresholds (10 and 20 MeV) with 10% uncertainties


Conversion based on altitude, Geomag. cut-off, and solar activity to estimate location dependence

of measurements

��36

Gordon et al. 2004

NIST vs Gran Sasso National Lab

Table 7.14: A comparison between the LNGS and FaNS-2 measurements of theneutron flux above 10 MeV and 20 MeV.

Energy Range LNGS Flux Corrected LNGS FaNS-2 Fluxn/cm2/s n/cm2/s n/cm2/s

En > 10 MeV (4.7±0.5)⇥ 10�3 (2.9±0.3)⇥ 10�3 (3.05± 0.4)⇥ 10�3

En > 20 MeV (4.2±0.4)⇥ 10�3 (2.6±0.3)⇥ 10�3 (2.75± 0.4)⇥ 10�3

During the course of six months, the detector was stable and operated ex-

tremely well. FaNS-2 has demonstrated the ability to measure very high energy

neutron events while still having sensitivity to low energy neutrons with good en-

ergy resolution. The comparison between FaNS-2 data and the MCNP simulation

is now discussed, along with a comparison between the two FaNS detectors.

7.8.1 Comparison to Monte Carlo

Though there are many fluctuations in measurements of the total flux of

cosmic-ray induced neutrons, the relative shape of the energy spectrum is sim-

ilar across measurements. Thus, the Monte Carlo has been used to determine

the weighted response (in neutron detected per neutron fluence) for FaNS-2 in the

cosmic-ray induced neutron field. This response has then been applied to the de-

tected neutron rate to obtain a measurement of the incident neutron flux.

Between 1 and 200 MeV the data and Monte Carlo spectra agree very well,

including features at 1 MeV and 3 MeV that appear in both spectra. These fea-

tures are part of the simulated spectrum because of calculations of resonances in

the atmosphere. However, Bonner sphere measurements lack the energy resolution

required to observe them. FaNS-2 has been able to directly observe these features.

241

10-8

10-7

10-6

10-5

10-4

10-3

10-2

10-1

Coun

ts /

MeV

/s

100 101 102 103 104

Energy (MeV)

Data MCNP

(MeV)visE10 210

)−

1flu

x (H

z M

eV

−310

−210

−110

1CGS Measurement

at LNGS


To compare between LNGS and NIST, a

conversion factor (x1.64) based on location, and

solar cycle has been applied

Bonardi et al. Astroparticle Phys 34 (2010) 225–229

��37

FaNS Outlook• Both FaNS-1 and FaNS-2 are operating at NIST

• FaNS-1 is characterizing fast neutron backgrounds at reactors for PROSPECT, a new neutrino oscillation experiment that will operate at sea-level

• FaNS-2 is measuring the muon-induced neutron spectrum at a depth of 20 m.w.e.

• Work is underway to better understand the deviation between data and MCNP for En > 200 MeV

• The FaNS program has definitively demonstrated the benefit of CGS for neutron detection

• The FaNS detectors are improving NIST’s functionality for precision neutron spectroscopy


Conclusions• The surface and underground fast neutron backgrounds pose

serious threats to the feasibility of current and future low-background experiments

• Current detection techniques lack the ability to fully characterize the spectrum and flux of these backgrounds

• Two novel detectors based on Capture Gated spectroscopy have been developed and calibrated by the UMD/NIST collaboration

• FaNS-1 successfully performed a measurement of the surface fast neutron spectrum before being deployed at KURF

• FaNS-1 characterized the fast neutron background at KURF demonstrating that the spectrum is (alpha,n)-like

• FaNS-2 greatly improved on the measurement of the surface fast neutron spectrum, and made a definitive measurement of the flux


the umd/nist fast neutron spectrometers

Documents