Experimental techniques for neutron detection

Experimental techniques for neutron detection

Roberto Bedogni

INFN-LNF Frascati

(L1) Introduction on neutron interaction and 2011 ICRU-recommended quantities •  Quantities and Units (ICRU 85) •  Neutron Interaction •  Examples and videogames

(L2) Neutron Measuring Instruments •  Slow Neutron Detectors •  Fast neutron detectors and spectrometers •  Dosemeters

(L3) Calibration fundamentals •  Calibration fundamental •  Workplace and calibration fields

(L4) Case study •  CS intro •  Neutron source term •  Instruments and measurement points

Experimental techniques for neutron detection

When the ith Bonner sphere is uniformly exposed in a point of a neutron field, the reading Ci of the central thermal neutron sensor can be expressed as Φ is the neutron fluence in cm-2;

Ri(E) is the response function of the sphere (in cm2). It is usually derived with Monte Carlo calculations and represents the reading per unit fluence as a function of the mono-energetic neutron energy, E. The set of response functions of all Bonner spheres forms the “response matrix”.

ϕ(E) is the energy distribution of the neutron fluence normalized to 1 cm-2 and its unit is MeV-1 (also termed “unit spectrum”).

The energy distribution of the neutron fluence (also termed “spectrum”), is given by Φ(E) = Φ ·ϕ(E) and its unit is cm-2·MeV-1. When a set of m Bonner spheres is exposed to the same neutron fluence, a set of readings Ci, i=1,..,m is collected.

∫=max

min

)()(E

Eii dEEERC ϕΦ

C: \ L2 \ Fast neutron detectors \ BSS

Response of BSS

The neutron fluence Φ and its energy distribution ϕ(E) are derived by inverting a set of m equations, that for computer calculation purposes can be expressed in the following discrete form:

Ng is the number of energy groups; Φ is the total neutron fluence in cm-2; Rij (cm2) is the response function of the i-th sphere in the j-th discrete energy

group; ϕj discrete energy distribution of the neutron fluence normalized to 1 cm-2

(MeV-1) (discrete version of the unit spectrum).

The energy distribution of the neutron fluence (discrete spectrum), is given by

Φj = Φ·ϕj (unit: cm-2·MeV-1)

ΔEj is the width of the j-th energy group (in MeV).

∑=

=gN

jjjiji ERC

1ΔϕΦ


Response of BSS

The response function of a sphere is usually derived by Monte Carlo simulations under broad parallel field geometry (see figure). The calculated response will be adequate for any direction distribution of the neutron fluence, provided uniform irradiation of the sphere.

The response function must be calculated for a significantly dense energy binning. The number of mono-chromatic energies to be simulated must be consistent with the complexity of the response function. Fine structures such as narrow resonances should be carefully taken into account

For the most popular central detectors, such as 6LiI(Eu), 3He or Au foils, from 50 to 100 simulations per sphere are needed to obtain an accurate approximation (few % in terms of overall uncertainty) of the response function of a given sphere. This should be increased if other materials with complex resonances, such as cadmium, are included in the spheres


Response matrix

Response function of a gold foils based BSS, using gold disks with diameter 1 cm and thickness 0.1 mm


Response of BSS

CALIBRATE a BSS means to verify, in a well known neutron field (known fluence rate and its energy distribution) - better if it is a ISO reference field- the ratio between the CALCULATED response matrix and the ACTUAL one.

This ratio can be quite different from one, but the necessary condition is that it remain constant for all the spheres.

In practice, this procedure corresponds to the determination of the “effective” number of atoms in the sensitive elements.

Calibration


A neutron calibration room should be constituted by a large room, in order to minimize the scattered radiation in the point of test. Typically, a very large room is 10 m x 10 m x 10 m with aluminum walls The scattered radiation is ALWAYS important in neutron irradiation.

The reference fluence rate (or uncollided fluence rate) Φref at a distance l from a point source is:

Where l is the source-to-centre detector distance, B the source strength, Fanis the anisotropy factor, Σ  the air attenuation coefficient, Σ  r is the sphere radius.

Calibration

Φuncollided =B

4 ⋅π ⋅ l2⋅Fanis ⋅e

−Σ⋅(l−r )


If the calibration room is not “large” (a realistic room, e.g. 10 m x 10 m x 3 m with concrete walls):

And the BSS reading will be Ctot = Cuncoll + C scattered The SCATTERING CORRECTION TECHNIQUES recommended in ISO Standard 8529-2 allow in principle to determine Cscatter. Thus, for every sphere: •  The unit spectrum of the reference field is given in ISO Standard 8529-1

(Tabulated) •  The uncollided fluence rate Φuncoll is known from source data and distance

Calibration

Φtot =Φuncoll +Φscattered =B

4 ⋅π ⋅ l2⋅Fanis ⋅e

−Σ⋅(l−r ) +Φscattered


Ctot =Φuncoll Ri (E)ϕuncoll (E)dE0

Emax

∫ +Φscattered Ri (E)ϕ scattered (E)dE0

Emax

∫

fi =Ci uncoll

Φuncoll ⋅ R(E)i ⋅ϕ j

REF FIELD dE∫

BSS calibration table, Am-Be source


Calibration

Irradiation of a BSS using the shadow-cone technique


Calibration

Validation of the response matrix

The accuracy of a BSS is strictly related to the accuracy of the response matrix. The experimental validation of the response matrix in well known neutron fields is therefore an essential requisite for reliable spectral evaluations.

The overall uncertainty of a response matrix may be estimated from validation experiments in reference fields with similar energy distribution to the fields to be measured.

The experiment should be (possibly) done in monoenergetic fields


R. Bedogni et al. Nuclear Inst. and Methods in Physics Research, A 897 (2018) 18–21

Table 2The main characteristics of the used beams.Nominal monoenergetic energy(keV)

Full width of energydistribution (keV)

Reaction used Measurement angle Standard uncertainty on referencefluence at reference point

Target scatter fraction (% of thetotal fluence)

143.3 ± 4.6 14.4 7Li(!, ") 0◦ ±2.3% 1.1%565.6 ± 3.4 9.6 7Li(!, ") 0◦ ±2.5% 0.6%1201 ± 11 100 T(!, ") 0◦ ±2.9% 3.6%

Fig. 3. Simulated response matrix of the LL-BSS. The experimental response#$(%) is given by &'$(%), where the & constant (the calibration factor) shouldnot depend on the neutron energy and sphere size.

As expected, the shape of the LL-BSS response matrix is very similarto that of other BSSs described in literature.

As not all neutron captures in the crystal yield a measured pulse,the experimental response of the LL-BSS unit fluence, #$(%), will belower than the corresponding simulated quantity '$(%). This can beattributed to multiple factors, such as the crystal-to-PM optical coupling,the difference between real and nominal size of the crystal, the exactcrystal composition.

The ratio between observed and expected counts is hereafter calledspectrometer calibration factor, & , and its value is lower than 1. Itsexperimental determination is one of the objectives of this work. Ifthe simulation model is correct in terms of geometry, dimensions andmaterials, & will not depend from the sphere size and from the neutronenergy.

For every monoenergetic energy used in this experiment, and forevery sphere, an experimental estimation of the calibration factor &$,%(where $ denotes the sphere and % the mono-chromatic energy) wasperformed. Its degree of constancy, when the sphere size and the neutronenergy vary, is here regarded as the ‘‘overall accuracy’’ of the simulatedresponse matrix in the studied energy range.

4. Irradiation conditions

The irradiation tests took place in the low-scatter irradiation roomof NPL, using the 3.5 MV Van de Graaff accelerator operated by theNeutron Metrology Group. The exploited reactions were 7Li(!, ") togenerate 144 keV and 565 keV beams, and ( (!, ") for the 1.2 MeV beam.All measurements were done at 0◦ angle with respect to the directionof the accelerated beam. The distance from target to the sphere centrewas 2 m. Eleven spheres (from 60 to 250 mm in diameter) were usedin the experiment. The shadow-cone technique was used to subtract theair- and room-scatter contribution from the spectrometer readings.

The reference value of monoenergetic neutron fluence delivered tothe reference point (sphere centre) was known through measurements

with the Standard NPL long counter plus a suite of permanent monitorinstruments.

The main characteristics of the used beams are summarized inTable 2.

The target scatter fraction, based on Monte Carlo simulations, can beassumed as affected by uncertainty up to ±40% [14]. These are neutronswith lower energy than the monoenergetic one. Since the long counterused to determine the reference fluence has flat energy dependenceof the fluence response, the result is a slight overestimation of themonoenergetic fluence. This was taken into account in data analysis.

5. Results of the irradiation tests

For every monoenergetic energy (%) and every sphere ($), an estima-tion of the LL-BSS calibration factor &$,% , was derived as follows.

If the beam was a pure monoenergetic one, &$,% could be simplyderived by the formula (1)

&$,% =

)$,%,*+*,$,%,*+*

− )$,%,)+"-,$,%,)+"-

'$,%(1)

where:

#$,%,*+* counts in the $th sphere in the total field irradiation;,$,%,*+* monoenergetic neutron fluence delivered at the reference

point (sphere centre) during the total field irradiation (unit:cm−2);

#$,%,)+"- the counts in the $th sphere in the irradiation with shadowcone;

,$,%,)+"- monoenergetic reference neutron fluence that would be deliv-ered at the reference point during the irradiation with shadowcone, in absence of shadow-cone (unit: cm−2);

'$,% simulated sphere response function at the energy %.

To be more precise, it should be considered that the target-scatteredcomponent is affecting both the value of fluence and the sphere counts,but not in the same way. Therefore, the fraction in Eq. (1) needs tobe multiplied by a correction factor, ./, elaborated on the basis of theexpected monoenergetic and target scattered spectra, provided by NPL../ is defined as follows:

./ =1 − (0 1/)2**-3

1/)2**-3+4+"+

1 − (0(2)

where:

TS Target-scatter fraction (Table 2) in terms of fraction oftotal the fluence;

1/)2**-3 Sphere response function folded with the scattered spec-trum normalized to unit fluence;

1/)2**-3+4+"+ Sphere response function folded with the total spectrumcoming from the target, i.e. monoenergetic + target scat-tered, normalized to unit fluence.

./ usually constitutes a small correction: its value is ranges from0.991 ± 0.014 (BS60) to 1.004 ± 0.007 (BS250) for 144 keV, from0.995 ± 0.008 (BS60) to 1.002 ± 0.004 (BS250) for 565 keV, and from0.933 ± 0.071 to 1.018 ± 0.019 for 1.2 MeV.

The &$,% values experimentally obtained are reported in Table 3 forevery monoenergetic energy

20




Table 3!",# values measured for every sphere and energy, and best estimation perenergy.

144 keV 565 keV 1.2 MeVBS60 0.695 ± 4.6% 0.684 ± 4.2% Not doneBS70 0.715 ± 4.1% 0.698 ± 3.6% 0.742 ± 8.3%BS80 0.703 ± 3.9% 0.710 ± 3.4% 0.734 ± 7.2%BS90 0.719 ± 3.8% 0.705 ± 3.2% 0.737 ± 6.4%BS100 0.708 ± 3.7% 0.707 ± 3.1% 0.727 ± 5.8%BS110 0.720 ± 3.6% 0.699 ± 3.1% 0.731 ± 5.5%BS125 0.685 ± 3.5% 0.682 ± 3.0% 0.709 ± 5.0%BS150 0.715 ± 3.4% 0.701 ± 2.9% 0.714 ± 4.5%BS170 0.710 ± 3.4% 0.700 ± 2.9% 0.722 ± 4.3%BS200 0.702 ± 3.4% 0.711 ± 2.8% 0.722 ± 4.0%BS250 0.657 ± 3.3% 0.651 ± 2.8% 0.726 ± 3.8%!# 0.701 ± 2.5% 0.694 ± 2.7% 0.724 ± 3.3%! = 0.703 ± 1.6%

Fig. 4. Comparing the experimental (points) and simulated (lines) responsefunction for the spheres from BS 60 to BS 100. Error bars refer to one sigma.

The best estimation of the calibration factor was derived for everyenergy, using the inverse square of uncertainty as weighting factors.This corresponds to !# in Table 3 Uncertainties of !# values are mainlydue to fluence uncertainties (from Table 2).

The global calibration factor, ! (last line of Table 3), was thenobtained by a weighted average of the !# values Its numerical valueis ! = 0.703 ± 0.011 (±1.6%). The ±1.6% figure can be regarded as anestimation of the ‘‘overall uncertainty’’ of the simulated response matrixfor the investigated energy range.

Figs. 4 and 5 compare, for the different spheres and energy, theexperimental response function ( $",#,%&%

'",#,%&%− $",#,$&()

'",#,$&()) *+ and the simulated

one (corrected with the calibration factor), ! ,",# . The comparison issatisfactory.

6. Conclusions

A Bonner Sphere spectrometer employing a ‘‘large’’ (11 mm diameter× 3 mm thickness) 6LiI(Eu) scintillator (LL-BSS), was developed andcalibrated in monoenergetic reference fields of 144 keV, 565 keVand 1.2 MeV at NPL. The central detector, directly coupled to a


digitizer, exhibits a very clean pulse height distribution with a verylarge separation from neutron to photon events. With respect to thetraditional BSS based on the 4 mm (diameter) × 4 mm (height) 6LiI(Eu),this new BSS is a factor of 3 more sensitive. An accurate MCNP model ofthe LL-BSS was used to generate a 120 equilethargic groups theoreticalresponse matrix (from 1E-9 to 1E+3 MeV). The measurements showedthat the overall uncertainty of this matrix in the studied energy rangeis lower than ±2%. Further work is planned to introduce metal-loadedspheres for high-energy neutrons, and to extend the range of validationto 10–20 MeV and higher energies.

Acknowledgements

This work has been supported by projects INFN-E andE_LIBANS from INFN (Commissione Scientifica Nazionale 5). The staffof the NPL Neutron metrology Group is greatly acknowledged.

References

[1] R.L. Bramblett, R.I. Ewing, T.W. Bonner, Nucl. Instrum. Methods Phys. Res. 9 (1960)1.

[2] A.V. Alevra, D.J. Thomas, Radiat. Prot. Dosim. 107 (2003) 33.[3] ISO 8529-1, International Organization for Standardization. Reference neutron ra-

diations e Part 1: characteristics and methods of production. International StandardISO 8529-1, 2001.

[4] D.J. Thomas, A.V. Alevra, Nucl. Instrum. Methods Phys. Res. A 476 (2002) 12.[5] R.M. Howell, E.A. Burgett, B. Wiegel, N.E. Hertel, Radiat. Meas. 45 (2010) 1233.[6] V. Mares, H. Schraube, Nucl. Instrum. Methods Phys. Res. A 337 (1994) 461.[7] V. Lacoste, V. Gressier, J.-L. Pochat, F. Fernandez, M. Bakali, T. Bouassoule, Radiat.

Prot. Dosim. 110 (2004) 529.[8] B. Wiegel, A.V. Alevra, Nucl. Instrum. Methods Phys. Res. A 476 (2002).[9] S. Barros, V. Mares, R. Bedogni, M. Reginatto, A. Esposito, I.F. Gonçalves, P. Vaz, W.

Rühm, Radiat. Prot. Dosim. 161 (2014) 46.[10] A. Pola, D. Bortot, M.V. Introini, R. Bedogni, A. Gentile, A. Esposito, J.M. Gómez-Ros,

E. Passoth, A. Prokofiev, Radiat. Prot. Dosim. 161 (2014) 229.[11] D.B. Pelowitz, MCNPX User’s Manual Version 2.7., Report LA-CP-11-00438, 2011.[12] M.B. Chadwick, P. Oblozinsky, M. Herman, Nucl. Data Sheets 107 (2006) 2931.[13] C. Pioch, V. Mares, W. Rühm, Radiat. Meas. 45 (2010) 1263.[14] National Physica Laboratory, NPL. Certificate of calibration N1438. 1st December

2016.

21

∑=

− ⋅⋅⋅

=g

j

N

jj

BeAmijuncoll

ii

ER

Cf

1ΔϕΦ




Table 3! ",# values measured for every sphere and energy, and best estimation perenergy.

144 keV 565 keV 1.2 MeVBS60 0.695 ± 4.6% 0.684 ± 4.2% Not doneBS70 0.715 ± 4.1% 0.698 ± 3.6% 0.742 ± 8.3%BS80 0.703 ± 3.9% 0.710 ± 3.4% 0.734 ± 7.2%BS90 0.719 ± 3.8% 0.705 ± 3.2% 0.737 ± 6.4%BS100 0.708 ± 3.7% 0.707 ± 3.1% 0.727 ± 5.8%BS110 0.720 ± 3.6% 0.699 ± 3.1% 0.731 ± 5.5%BS125 0.685 ± 3.5% 0.682 ± 3.0% 0.709 ± 5.0%BS150 0.715 ± 3.4% 0.701 ± 2.9% 0.714 ± 4.5%BS170 0.710 ± 3.4% 0.700 ± 2.9% 0.722 ± 4.3%BS200 0.702 ± 3.4% 0.711 ± 2.8% 0.722 ± 4.0%BS250 0.657 ± 3.3% 0.651 ± 2.8% 0.726 ± 3.8%! # 0.701 ± 2.5% 0.694 ± 2.7% 0.724 ± 3.3%! = 0.703 ± 1.6%


The best estimation of the calibration factor was derived for everyenergy, using the inverse square of uncertainty as weighting factors.This corresponds to ! # in Table 3 Uncertainties of ! # values are mainlydue to fluence uncertainties (from Table 2).

The global calibration factor, ! (last line of Table 3), was thenobtained by a weighted average of the ! # values Its numerical valueis ! = 0.703 ± 0.011 (±1.6%). The ±1.6% figure can be regarded as anestimation of the ‘‘overall uncertainty’’ of the simulated response matrixfor the investigated energy range.

Figs. 4 and 5 compare, for the different spheres and energy, theexperimental response function ( $",# ,%&%' ",# ,%&%

! $",# ,$&( )' ",# ,$&( )

) * + and the simulatedone (corrected with the calibration factor), ! , ",# . The comparison issatisfactory.

6. Conclusions

A Bonner Sphere spectrometer employing a ‘‘large’’ (11 mm diameter" 3 mm thickness) 6LiI(Eu) scintillator (LL-BSS), was developed andcalibrated in monoenergetic reference fields of 144 keV, 565 keVand 1.2 MeV at NPL. The central detector, directly coupled to a


digitizer, exhibits a very clean pulse height distribution with a verylarge separation from neutron to photon events. With respect to thetraditional BSS based on the 4 mm (diameter) " 4 mm (height) 6LiI(Eu),this new BSS is a factor of 3 more sensitive. An accurate MCNP model ofthe LL-BSS was used to generate a 120 equilethargic groups theoreticalresponse matrix (from 1E-9 to 1E+3 MeV). The measurements showedthat the overall uncertainty of this matrix in the studied energy rangeis lower than ±2%. Further work is planned to introduce metal-loadedspheres for high-energy neutrons, and to extend the range of validationto 10–20 MeV and higher energies.

Acknowledgements


References








2016.

21


Table 3! ",# values measured for every sphere and energy, and best estimation perenergy.

144 keV 565 keV 1.2 MeVBS60 0.695 ± 4.6% 0.684 ± 4.2% Not doneBS70 0.715 ± 4.1% 0.698 ± 3.6% 0.742 ± 8.3%BS80 0.703 ± 3.9% 0.710 ± 3.4% 0.734 ± 7.2%BS90 0.719 ± 3.8% 0.705 ± 3.2% 0.737 ± 6.4%BS100 0.708 ± 3.7% 0.707 ± 3.1% 0.727 ± 5.8%BS110 0.720 ± 3.6% 0.699 ± 3.1% 0.731 ± 5.5%BS125 0.685 ± 3.5% 0.682 ± 3.0% 0.709 ± 5.0%BS150 0.715 ± 3.4% 0.701 ± 2.9% 0.714 ± 4.5%BS170 0.710 ± 3.4% 0.700 ± 2.9% 0.722 ± 4.3%BS200 0.702 ± 3.4% 0.711 ± 2.8% 0.722 ± 4.0%BS250 0.657 ± 3.3% 0.651 ± 2.8% 0.726 ± 3.8%! # 0.701 ± 2.5% 0.694 ± 2.7% 0.724 ± 3.3%! = 0.703 ± 1.6%


The best estimation of the calibration factor was derived for everyenergy, using the inverse square of uncertainty as weighting factors.This corresponds to ! # in Table 3 Uncertainties of ! # values are mainlydue to fluence uncertainties (from Table 2).

The global calibration factor, ! (last line of Table 3), was thenobtained by a weighted average of the ! # values Its numerical valueis ! = 0.703 ± 0.011 (±1.6%). The ±1.6% figure can be regarded as anestimation of the ‘‘overall uncertainty’’ of the simulated response matrixfor the investigated energy range.

Figs. 4 and 5 compare, for the different spheres and energy, theexperimental response function ( $",# ,%&%' ",# ,%&%

! $",# ,$&( )' ",# ,$&( )

) * + and the simulatedone (corrected with the calibration factor), ! , ",# . The comparison issatisfactory.

6. Conclusions

A Bonner Sphere spectrometer employing a ‘‘large’’ (11 mm diameter" 3 mm thickness) 6LiI(Eu) scintillator (LL-BSS), was developed andcalibrated in monoenergetic reference fields of 144 keV, 565 keVand 1.2 MeV at NPL. The central detector, directly coupled to a


digitizer, exhibits a very clean pulse height distribution with a verylarge separation from neutron to photon events. With respect to thetraditional BSS based on the 4 mm (diameter) " 4 mm (height) 6LiI(Eu),this new BSS is a factor of 3 more sensitive. An accurate MCNP model ofthe LL-BSS was used to generate a 120 equilethargic groups theoreticalresponse matrix (from 1E-9 to 1E+3 MeV). The measurements showedthat the overall uncertainty of this matrix in the studied energy rangeis lower than ±2%. Further work is planned to introduce metal-loadedspheres for high-energy neutrons, and to extend the range of validationto 10–20 MeV and higher energies.

Acknowledgements


References








2016.

21

How many spheres?

Ideal (left) vs real (right) response matrix of a BSS


For operational purposes, an approach due to Alevra (NIM A 476, 12-20), valid for pure polyethylene spheres, may be used. See Figure for its application to a BSS equipped with a 4 mm x 4 mm 6LiI(Eu) counter. The counts of different spheres exposed to the same fluence of 241Am-Be or 252Cf are reported in function of the sphere diameter. The curves are fitted with a 4th order polynomial curve, meaning that five well chosen spheres are enough to describe the variability of the BS counts for all possible sphere diameters. For high-energy fields, where metal loaded spheres are used, this rule can not be applied.

Investigate the rank of the

structure matrix TR·R


How many spheres?

The response matrix needs to be validated with exposures in well known neutron fields. Although a complete covariance matrix is not available, the overall uncertainty of the response matrix in a given energy interval may be estimated from validation experiments. The following sources of uncertainty should be considered: •  The geometry and materials used for the Monte Carlo simulations.

•  The relative change in the sphere response due to change in the polyethylene density, (dR/R)/(dρ/ρ), is a function of the neutron energy and the sphere diameter.

This quotient, which can be as high as 6, is more important in energy regions where the sphere response is poor. A method for correcting the response function for changes in density and diameter is reported in Radiat. Prot. Dosim. 107(1-3), 37-72 (2003).

•  The difference between the simulated irradiation geometry and the real one. If the sphere is not uniformly irradiated, its response function may vary significantly

•  Isotropy of the central detector. For spherical detectors this aspect can be neglected. For directional detectors such as large gold foils, experimental or computational studies are

needed to evaluate the influence of the irradiation geometry on the detector response (Radiat. Prot. Dosim. 128 (3) 89–293 (2008)).

•  Stability, dose rate dependence and photon sensitivity of the central detector


Uncertainties

- The weight;

-  The poor energy resolution: fine spectral structures are ignored, unless the unfolding code is specifically instructed to detect them;

- The need to sequentially irradiate the spheres, requiring long irradiation sessions.

-  Another experimental difficulty, especially for big spheres, is the need to uniformly irradiate the spectrometer, which is the condition under which the response functions are calculated. This condition can be fulfilled with point sources at large distances. In collimated beams as from accelerators, beam-scanning techniques have to be implemented.

-  It should be noticed that the “uniform irradiation condition” does not necessarily imply an aligned expanded field: the angular distribution of the field can be whatever, since the sphere response is isotropic.

-  Unfolding codes are usually very complex, requiring detailed pre-information (such as a default spectrum as close as possible to the real one) and expert users.

- Complex analysis to get uncertainties.

Complexity of BSS spectrometry


Among the available neutron spectrometry techniques, the BSs is the most frequently used in radiation protection.

D. J. Thomas. Neutron spectrometry for radiation protection. Radiat. Prot. Dosim. (2004), 110 (1-4), 141-149.


Single Moderator Spectrometers SMS

�  S imul taneous deter minat ion of a l l energy

components (eV - hundreds MeV) of a neutron field using a single-moderator instrument

�  Single moderator embedding multiple thermal neutron detectors

�  Different dephts � different E(max response), like BSS Isotropic response: SP2 Directional response: beam diagnostics, cosmic

�  Simple operation and BSS-like environment

C: \ L2 \ Fast neutron detectors \ SMS

BSS is still unsurpassed for workplace measurements on large energy range (eV to hundreds of MeV)

but •  Time consuming irradiation sessions •  Unsuited for real-time monitoring •  poor energy resolution (partially compensated with

a-priori information) •  Unfolding requires specific skills


SP2

SPherical SPectrometer

•  Spherical geometry to match isotropic response

•  31 detectors combined as a function of the radius

•  Efficient n/g discrimination •  Internal Pb shell for high-E

•  Mimics a 6-spheres BSS

•  RP sector

CYSP

CYlindrical SPectrometer

•  Cylindrical collimated geometry •  7 detectors at different depths •  Efficient n/g discrimination •  Internal Pb shell for high-E

•  Mimics a 7-spheres BSS •  Accelerator, cosmic ray, HS


Internal detectors

Suitable thermal neutron detectors were developed by covering commercial semiconductor devices with 6LiF (1) Small size (1 x 1 x 0.4 cm) (2) Adequate Sensitivity (3) Good photon rejection (4) Low-cost (≈ 102 €)

- Multi-detector Low-V analog board developed - Labview-based DSP

For Peer Review

Pulse height distribution measured from the new active thermal neutron detector with an active layer of 6LiF irradiated with thermal neutrons in the ex-core radial column of the TRIGA reactor.

127x101mm (300 x 300 DPI)

Page 12 of 13

http://www.rpd.oupjournals.org

Radiation Protection Dosimetry Submitted Manuscript


SP2 design

• Thirty-one thermal neutron detectors symmetrically allocated along three axes of a 30 cm HDPE sphere

• Internal one cm thick lead shell (3.5 to 4.5 cm in radius) to enhance high-E response

• Positions: radius 0.0 (centre), 5.5, 7.5, 9.5, 11 and 14.5 cm (external)

R. Bedogni et al, NIM A 613, 127-133 (2010) J.M. Gomez-Ros et al, NIM A 677, 4-9 (2012)

• Isotropic response achieved by averaging the detectors located in the same radial position

(Maximum inaccuracy few % in eV-keV region for superficial positions)


SP2 simulated response

1 0 0 ISO

1 1 1


SP2 experimental response

-  2012. Measurements at PTB with mono-energetic neutron fields 147 keV - 14.8 MeV using a passive prototype (Dy activation foils) - 3% overall uncertainty

R. Bedogni et al, NIM A 714 (2013) 110-114. -  2013-2014. First tests of active device (detectors along one

radius) with Am-Be source + shadow cones to investigate response Vs. irradiation geometry - 2% overall uncertainty

R. Bedogni et al, NIM A 767 (2014) 159-162.


SP2 experimental response

October 2015. Testing the final instrument (31-active-detectors) with an Am-Be source.

Typ. response one cps / microSv/h


• Seven TNDs along the axis • Spectral resolution and lateral rejection • HPDE Collimator 50 cm diam x 30 cm h

Hole diameter 16 cm, B-plastic lined • 35 cm h x 50 cm diam detectors part • Capsule for detectors: 20 cm diam,

includes one cm lead disk (high-E) • Air holes to increase deep response

HDPEAIRPbB-plas0c

CYSP design

Radiat. Meas. 82, 47-51 (2015)


The CYSP spectrometer

Different pieces and assembling stages

Collimator (left) and detecting part (right)

Half detecting part with detector capsule

CYSP assembled + EM shield


CYSP energy response

2013: Mono-energetic neutron fields from 144 keV to 16.5 MeV at NPL (UK)

Overall uncertainty of Response matrix estimated as < ±2% (comparison between observed and calculated count rates)

NIM A 782 (2015) 35-39


CYSP directional response



Cosmic neutron field measurements with CYSP (HS version)

Zugspitze Mountain (2650 m)

1.  Scientific motivation

2.  The Zugspitze station and the HMGU-ERBSS

3.  The CYSP-HS spectrometer

5.  Comparing ERBSS and CYSP spectra

6.  Conclusions and perspectives

1.  Need for real time measurement of neutrons from cosmic rays at high elevations

� Benchmarking codes for aircrew dosimetry, e.g. EPCARD � Prompt detection of field anomalies potentially leading to aircrew overexposures

2.  www.nmdb.eu: world network of ≈ 50 neutron counters, but only one spectrometer: the HMGU-ERBSS at Zugspitze mt. Disturbing effects: Albedo, surrounding materials, snow.

3.  Idea behind this work: evaluate if the directional spectrometer CYSP (INFN-NEURAPID project) can be used with similar purposes:

� Measuring the high-E component, mainly downwards, disregarding contributions from other directions (albedo).

� With respect to BSS: more compact and less expensive

CYSP installed in March 2016 on top of Zugspitze !


The HMGU neutron measurement station at UFS Schneefernerhaus (2650 mt, 4.1 GV)

1.  He-3 p.c. based Bonner Sphere Spectrometer with 15 spheres, 2 of which with extended range + 1 bare detector.

2.  Continuous measurements (time resolution 1 h) since 2004 (Since Oct. 2005 in instrumental shed).

Photo: V. Mares 2016


The CYSP-HS internal detectors

� CYSP’s internal detectors (LATNDs) are sandwiches of two 2.8 x 2.8 cm2 Silicon diodes, “watching” the same 6LiF radiator

Bare diode Deposited diode Assembled LATNDs

�  8-channels custom analogue board (PoliMi)

�  Every channel has charge pre-amplifier and shaper amplifier

�  A high-speed digitizer (tens MS/s) processes the signals:

- EM interferences elimination - Non-thermal-neutron-events subtraction (using the PHD of non deposited LATND)


The CYSP-HS spectrometer Response matrix


Neutron spectra 2016, Week 26, average spectra

All data are pressure-corrected


High-E fluence rate

0.022 cm-2s-1

C: \ L2 \ Fast neutron detectors \ unfolding

Spectral unfolding in moderator-based instruments

BSS & SMS

1.  Guess-error model

2.  Ways to account for pre-information

3.  FRUIT

4. Examples

The unfolding procedure is the most complex task in BSS/SMS spectrometry

Unfolding algorithms usually operate as in the diagram. The procedure starts from a trial spectrum (or default spectrum), that can be generated in different ways. The BSS/SMS response matrix is then applied to it, obtaining a set of “calculated” sphere counts. The comparison between the “calculated” and the “measured” counts provides an “error signal” that gives information about the vicinity of the trial s p e c t r u m t o t h e u n k n o w n f l u e n c e spectrum . The tr ial s p e c t r u m i s t h e n iteratively varied until the error signal is “small enough”, compared with t h e u n c e r t a i n t i e s . Different unfolding codes adopt different methods to generate and perturb the trial spectrum.


Unfolding

The unfolding problem with BSS is largely under-determined, because the number of readings (independent measurements), is always lower than the number of energy groups used for the representation of the spectrum (unknowns).

Of the infinite functions that could mathematically satisfy the problem, only a limited number is physically acceptable.

To restrict the whole space of functions to a physically meaningful neighborhood of the “solution”, the unfolding codes must be supplied with some pre-information about the spectrum to be measured.

-  A “default spectrum” as close as possible to the spectrum to be obtained is required in some unfolding codes such as MAXED (Health Phys. 77, 579 (1999)).

- Parametric codes, such as NUBAY (RPD 125 (1–4) 304–308 (2007)) and FRUIT (NIM A 580 1301-1309 (2007)), eliminate the non-physical solutions by modeling the neutron spectrum as a superposition of elementary spectra parameterized in terms of a small number (about ten) of physically meaningful parameters.

Unfolding


In a review of neutron spectrometry for radiation protection (RPD 110 (1–4), 141-149 (2004)), Thomas reported that the computer unfolding codes are almost invariably affected by the following:

-  the complexity of the codes -  the need of realistic a priori information to derive physically acceptable

solutions (such as a default spectrum as close as possible to the final result, usually derived with complex Monte Carlo simulations that need to accurately study the irradiation set up, the source composition, etc..),

-  need expert users


Unfolding (status 2004)

BSS/SMS are used in a large variety of neutron fields (nuclear plants, medical and research accelerators, radionuclide based facilities, calibration fields, cosmic rays) covering an energy range from thermal up to hundreds MeV. Modern unfolding codes, coupled with a minimal knowledge of the radiation environment, allow:

- limiting the impact of the under-determined nature of the problem; - treating the uncertainties;

Typical achievable accuracy: <5% for fluence or <10% for dose equivalent. BSS/SMS are regarded as the most adequate for most practical applications


Unfolding (current status)


Unfolding

-  The limited amount of a priori information needed to obtain meaningful solutions. Besides the sphere response functions, the counts and related uncertainties, the user is only asked to introduce qualitative information on the type of “radiation environment”, on the basis of a check-box input section.

-  No “default spectrum” is asked to the user. The code generates itself a default spectrum, needed to start the iterative procedure, on the basis of the radiation environment selected by the user.

-  Taking advantage of a “flexible tolerance” convergence mechanism, the results don’t depend on the numerical values of this initial spectrum.

-  User-friendliness and visual operation. The quantities involved in the unfolding process and their variation are continuously displayed: the plot of the spectrum, the measured and unfolded Bonner sphere counts, the parameters, the tolerances and the dosimetric quantities.

- Uncertainties of input quantities (sphere counts and response matrix) are propagated on the resulting integral quantities (fluence and dose equivalent).


Unfolding

Workplace measurements with BSS

Workplace: e+/e- collider; 510 MeV/beam Type of detector: 4x4 6LiI(Eu) active scintillator Spheres: 2”, 3”, 5”, 8”, 10”, 12”, 12”+ 1 cm Pb Irradiation mode: 20 min per sphere + rem counter for normalization


Workplace: 18 MV medical LINAC Type of detector: gold foils (10 mm diameter x 0.1 mm) Spheres: 2”, 3”, 5”, 8”, 10”, 12” Irradiation mode: 10 Gy at the isocentre per sphere

Workplace measurements with BSS


SMS Unfolding exercises

•  Focus on CYSP-NCT

•  Parametric unfolding & pre-information

1.  Spectrum from Am-Be source 2.  Parametric unfolding

•  SGM unfolding & pre-information

1.  Epithermal spectrum from E_LIBANS facility 2.  Unfold from Flat-in-lethargy guess (E > 0.1 eV) 3.  Unfold from realistic guess


SMS Unfolding exercises


Elastic scattering-based detectors

•  Elastic scattering from light nuclei, convenient because maximum recoil energy will be

•  In case of Hydrogen ER max = En

4-Helium ER max = 0.64 En Carbon ER max = 0.28 En

•  Very important simplification: isotropic scattering in CM (True for hydrogen

< 10 MeV). The expected proton recoil energy distribution is therefore a simple rectangle.

•  This theoretical shape in real H-scattering detectors is distorted by a number of effects.

•  Efficiency is ε = 1-exp(-Σ x)

C: \ L2 \ Fast neutron detectors \ scattering-based

Cross sections

C: \ L2 \ Fast neutron detectors \ scattering-based

Scintillator based recoil proton spectrometers

1.  Hydrogenated organic scintillators

2.  Higher efficiency than gas counters

3.  Proton range << detector size, no wall-effects

4.  Respond to photons (rejection methods)

5.  PHD does not differ too much from ideal rectangular shape

6.  Reasonably isotropic response

C: \ L2 \ Fast neutron detectors \ scintillators

Types of Scintillators for n spectrometry > 1 MeV •  Organic crystals

Delicate Anisotropic Expensive Comparatively small Antracene: high light yield Stylbene: better gamma rejection

•  Plastic Very easy to machine and cheap Good transparency Gamma discrimination is difficult

•  Liquid Better g/n discrimination Light output non linear with proton energy


NE213/BC501A LIQUID SCINTILLATION SPECTROMETERS RPD 107 (1–3) pp. 95-109 (2003)

High efficiency is obtained, compared with gas counters, due to the relatively high density of the scintillator. The fluorescence of stilbene crystals and that of NE213 (Nuclear Enterprise Ltd, Edinburgh, UK)/BC501A (Bicron Radiation Measurement Products, Newbury, OH, USA) depends on the stopping power of the secondary charged particles produced. The fraction of events with LONG lifetime increases as the stopping power increases. It is therefore possible discriminating between the neutrons and the photons, since their secondary have different stopping power (pulse shape analysis). For neutrons up to 20 MeV the cylindrical geometry 2” x 2” is an appropriate choice. The response functions are very well known and the response is nearly isotropic.


NE213/BC501A LIQUID SCINTILLATION SPECTROMETERS

Configuration of a liquid scintillator Pulse shape discrimination



Light output functions (LOF)



Response matrix (NRESP code) ��dN/dL for monoenergetic neutrons


Shape of response functions

LOF curves L(E) (both in MeV) are not linear but L(E) ≈ E3/2

In case of monoenergetic neutrons the spectrum dN/dL is the response function dN/dE = costant (rectangular) dN/dL = dN/dE / (dL/dE)

≈ E-1/2

≈ L -1/3

•  Wall effect in small volumes increases the slope at low-E


Calibration using gammas •  Electrons have larger LOF than protons (2-3 MeV protons

equivalent to 1 MeV electron)

•  LOF for electrons is fairly linear

•  Photons are used to calibrate the scintillator

•  No photoelectric peak, Compton edge is the only reference


Typical Performance •  Liquid scintillator Irradiated with 14 MeV monoenergetic


Gas-based recoil proportional counter (keV to 1 MeV)

1.  Filling gas can be H2, CH4

2.  For H, ideal response to monoenergetic neutrons is a rectangle, but different effects complicate the shape. Elastic scattering from 4-helium is not isotropic in the center-of-mass system, thus distributions can’t be rectangular.

3.  Efficiency << than scintillators

4.  Less linearity problems than scintillators

5.  Wall effects

6.  PSD useful to separate 2ry electrons from neutrons (recoil tracks are shorter than electron tracks)

7.  No multiple scattering

C: \ L2 \ Fast neutron detectors \ Gaseous counters

Proportional counter spectrometers (~ keV up to 1.5 MeV) RPD 107 (1–3) pp. 73–93 (2003)

Neutrons can be detected only indirectly by nuclear reactions where charged particles are produced. Two kinds of reactions can be used in gas-filled proportional counters: (1) Exothermic nuclear reactions resulting in secondary charged particles, e.g.

3He(n,p)3H + Q (Q = 764 keV)

A neutron of energy En produces a pulse of height En + Q. (2) Elastic scattering of neutrons with nuclei of the filling gas. This produces recoil protons in hydrogen (or a hydrogen-containing gas such as CH4) or alpha particles in 4He filling. The maximum energy transferred (for head-on collision) to the recoil nucleus is given by The energy distribution of the recoil nuclei ranges in energy from 0 to Emax and is in principle constant (isotropic scattering)


Proportional counter spectrometers (~ keV up to 1.5 MeV)

Proportional counters can be manufactured in right cylindrical or spherical geometries.

(1) Spherical counters have isotropic response. Skill is needed to produce high-resolution devices. Critical points are the purity of the filling gas and the constancy of the electric field over the whole anode wire.

(2) Cylindrical counters may be easily manufactured even in large dimensions, allowing extending the maximum energy of the response up to more than 4-5 MeV. Excellent electric field conditions are easy to achieve. The response is not isotropic.

The use of the elastic (n,p) reaction in H2 and CH4 filled recoil proton proportional counters:

•  Xs well known •  the scattering is isotropic in the centre-of-mass system up to 5 MeV •  the absolute value of cross section is reasonably high •  the mean energy loss of secondary protons is constant (36 eV per ion

pair) •  Energy calibration is difficult (internal source or traces 3He)


Proportional counter spectrometers (~ keV up to 1.5 MeV)

CH4 counters: higher stopping power for recoil protons (above 250 keV the tracks of recoil protons are more than 3.5 times shorter than in hydrogen), so that higher neutron energy may be measured. The production of carbon recoils is a disadvantage, because the maximum energy transferred to a carbon nucleus is 28% and because the ionization of carbon nuclei is 75% if compared with protons. The distribution of carbon recoils ends at 21% of the corresponding neutron energy. This must be taken into account in the determination of the response matrix.

Proportional counters filled with 4He extend the highest measurable neutron energy up to > 10 MeV, because the recoil nuclei (alpha particles) have shorter track lengths in the counter volume compared with recoil protons.

3He-filled proportional counters: exothermic 3He(n,p)T reaction. Pulse height spectrum reflects closely the shape of the neutron spectrum until about 1 MeV (where elastic scattering becomes important) Thermal neutrons appear in the spectrum in channel 764 keV. Photon discrimination is therefore easy (unless the gamma background is so high to cause pile-up).


SP2 Proportional counters (R= 2 cm, various pressure)

The most commonly used recoil counters are the spherical SP2 counters: •  high energy resolution (ΔE/E in the order of a few per cent) for neutron spectrometry. •  isotropic response because of their spherical shape. •  well known and experimented •  they can cover the 50–1500 keV energy range where fluence to dose equivalent coefficients vary rapidly with energy.

Disadvantages: •  Microphonic. To be enclosed in metal boxes. • .Since two or three counters are usually to be used in succession, long irradiation sessions are required to cover the whole energy range •  relatively low efficiency due to the low nuclear density of the gas filling compared with that of solid or liquid organic scintillators.


Response function of a SP2 counter at pressure 400 kPa

The number of ions collected at the anode is not fully proportional proton recoil energy because: -  not all recoil protons lose their entire energy within the counter before hitting the wall (wall distortion effects) -  gas amplification is not constant over the entire volume, as the electric field strength drops at the ends of the anode wire (gas amplification distortion effects).


SP2 Proportional counters

Energy range of SP2 at different pressures

Spectrometry with proportional counters can be extended to neutron energies higher than 1.5 MeV by using very large spherical counters filled with hydrogen, methane or propane and, in addition, Ar or Xe for the reduction of wall effects (up to 5 MeV) or if 4He filled counters are added.


Exercise SP2 (radius 2 cm) spherical H2 counter, 400 kPa, compare l i terature response functions with videogame (wall effect but not field distortion)


Exercise SP2, spherical H2 counter, 100 kPa, compare literature response function (144 keV) with videogame (GEB, Wall effect)

C: \ L2 \ Fast neutron detectors \ Gaseous counters H. TAGZIRIA and W. HANSEN

82

rays induce secondary electrons by interaction in thestainless steel wall of the detector. These electrons ion-ise the filling gas too, produce pulses at the detectoroutput and contribute to the measured spectrum. But,compared with recoil protons, the ionisation density (orthe energy loss per cm) for gamma induced electrons ismuch smaller, resulting in much longer track lengths forelectrons than for protons of the same energy.

From this behaviour two important conclusions canbe drawn:

Measured

Calculated

60 80 100 120 140 160 180

2500

2000

1000

1500

500

0

Energy (keV)

Cou

nts

Figure 4.9. Measured response of the 100 kPa SP2 counter to 144 keV monoenergetic neutrons compared with calculations.

24

22

20

18

16

14

12

10

8

60 200 400 600 800 1000

Energy (keV)

FW

HM

/E (

%)

Model: sqrt (A+B*E+C*E^2)/E

A = 66.83 ± 0.24B = 0.296 ± 0.002C = 0.0050 ± 0.0001

Figure 4.10. Relative FWHM pulse-height resolution of a 300 kPa (3 atm) SP2 counter as a function of energy.

(1) Gamma-induced electrons passing completelythrough the detector can lose only a certainmaximum energy in the counter volume, which isdetermined by the detector dimensions and thepressure and/or quality of the filling gas. Above thisenergy no gamma-induced pulses are observed.Hence, pulse-height discrimination is a simplemethod of n/! discrimination, i.e. the measuredpulse-height distribution is evaluated only abovethis maximum gamma energy that can be deposited.

at Florida Atlantic U

niversity on March 22, 2016

http://rpd.oxfordjournals.org/D

ownloaded from

Exercise 1.5 MPa atm , 6 cm radius spherical He4 compare literature response function with videogame


Recoil proton telescopes (keV to 1 MeV)

1.  Allow selecting a sharp range of recoil energy (in scintillators

or prop counters all are recorded)


Recoil proton telescopes (> 102 keV)

•  Incident n direction must be FIXED and well known

•  Use a polymer radiator with thickness << lowest recoil energy to be measured

•  In vacuum

•  Usually θ ≈ 0 (not zero to avoid exposing the detector in the incident beam)

•  Two detectors in coincidence (usually semiconductors) are used to reduce background (Ep = E1 + E2)

•  Very low efficiency (typ. 1E-5): - thin PE radiators have 1E-3 / 1E-4 yield; - the solid angle must be kept low, to favour E resolution

•  No wall effects, No multiple scattering

•  Active radiator can increase efficiency


Capture-gated spectrometers (> 102 keV)


Experimental techniques for neutron detection

Documents

Transcript of Experimental techniques for neutron detection