Content uploaded by Donatella Pascoli
Author content
All content in this area was uploaded by Donatella Pascoli on Nov 10, 2017
Content may be subject to copyright.
*Corresponding author. E-mail: Flavio.Cavanna@aquila.infn.it.
Nuclear Instruments and Methods in Physics Research A 418 (1998) 285—299
Calibration of BC501A liquid scintillator cells with monochromatic
neutron beams
The ICARUS Collaboration
F. Arneodo, P. Benetti, A. Bettini, A. Borio di Tigliole, E. Calligarich, C. Carpanese,
F. Casagrande, D. Cavalli, F. Cavanna*, P. Cennini, S. Centro, A. Cesana, C. Chen,
Y.B. Chen, D. Cline, O. Consorte, I. De Mitri, R. Dolfini, A. Ferrari, A. Gigli
Berzolari, K.L. He, X.P. Huang, Z.H. Li,F.Lu, J.M. Ma, G. Mannocchi, C. Matthey,
F. Mauri, L. Mazzone, C. Montanari, R. Nardo` , S. Otwinowski, S. Parlati,
D. Pascoli, A. Pepato, L. Periale, G. Piano Mortari, A. Piazzoli, P. Picchi,
F. Pietropaolo, A. Rappoldi, G.L. Raselli, S. Resconi, J.P. Revol, M. Rossella,
C. Rossi, C. Rubbia, P. Sala, D. Scannicchio, F. Sergiampietri, S. Suzuki, M. Terrani,
P. Torre, S. Ventura, M. Verdecchia, C. Vignoli, G.F. Xu, Z.Q. Xu, H. Wang,J.Woo,
C. Zhang, Q.J. Zhang, S.C. Zheng
Dipartimento di Fisica dell+Universita%dell+Aquila e INFN (LNGS), Via Vetoio, I-67010 L+Aquila, Italy
Dipartimento di Fisica dell+Universita%di Pavia e INFN (PV), Via Bassi 6, I-27100 Pavia, Italy
Dipartimento di Fisica dell+Universita%di Padova e INFN (PD), Via Marzolo 8, I-35131 Padova, Italy
CESNEF Politecnico di Milano e INFN (MI), Via Ponzio 34/3, I-20123 Milano, Italy
INFN Laboratori Nazionali di Frascati, Via E. Fermi 40, I-00044 Frascati (Roma), Italy
Dipartimento di Fisica dell+Universita%di Milano e INFN (MI), Via Celoria 16, I-20133 Milano, Italy
CERN, CH-1211 Geneva 23, Switzerland
Department of Physics UCLA, Los Angeles, CA 90024, USA
ICGF del CNR, C.so Fiume 4, I-10133 Torino, Italy
IHEP, 19 Yuquan Road, Shijingshan District, 100 039 Beijing, China
Received 19 March 1998
Abstract
The recoil proton energy response has been measured by exposing cylindrical cells, filled with BC501A BICRON liquid
scintillator, to mono-energetic neutron reference fields. We determine the required calibration parameters and report the
detailed procedures for the experimental data handling. A dedicated Monte Carlo simulation of the detector response and
efficiency has been performed. It showed good agreement with the measured quantities. The results from this calibration are
necessary for a detailed study of the neutron spectrum at the underground Gran Sasso Laboratory, with a neutron detector
made of 32 liquid scintillator cells, like those used during the calibration. 1998 Elsevier Science B.V. All rights reserved.
0168-9002/98/$19.00 1998 Elsevier Science B.V. All rights reserved.
PII: S 0 1 6 8 - 9 0 0 2 ( 9 8 ) 0 0 6 7 9 - 2
For a complete set of references and documents, see the
URL page: http://www.aquila.infn.it/icarus.
1. Introduction
A neutron spectrometer is operational at the
underground Gran Sasso Laboratory. The appar-
atus consists of 32 cylindrical cells (1 l each) filled
with hydrogenated liquid scintillator, BICRON
BC501A. It has been designed to perform flux and
energy spectra measurements of neutrons from
natural radioactivity in Hall C (mainly U and Th in
the rocks surrounding the detector and relative
radioactive decay chains at equilibrium). This in-
formation is necessary to evaluate precisely the
neutron background for the solar neutrino experi-
ment envisaged with the ICARUS 600 t detector
(i.e. to determine the energy spectrum of back-
ground electrons produced by photons following
neutron capture in liquid argon, in the solar neu-
trino energy range, from about 2 to 15 MeV).
The liquid scintillator technique was chosen for
its good particle (neutron to gamma) discrimina-
tion capability. The energy deposition in scintil-
lator materials is converted into light output. In an
organic scintillator the corresponding scintillation
pulse may be described satisfactorily by a sum of at
least two exponential time-dependent components
(“fast”and “slow”). The two components are
weighted differently depending on the excitation
induced by particles of different types. Neutrons
interact with the scintillating medium through
the process of proton recoil while gamma-rays
are primarily detected through Compton electron
recoil. The scintillation response for electron
ionization is faster than the response for proton
ionization. This is commonly used as a means of
suppressing gamma-ray background in neutron de-
tection systems.
The BC501A liquid scintillator characteristics
(equivalent to NE-213, produced by Nuclear Enter-
prises) are well known. In particular, it has been
selected for its enhanced emission of delayed light
(slow component) which allows optimal pulse-shape
discrimination (PSD), suitable for neutron spectro-
scopy in an intense gamma radiation background.
The aim of this work is the calibration of liquid
scintillator cells, prototype of the final neutron
spectrometer cells, by using monochromatic neu-
tron reference fields, in the energy range between
0.5 and 20 MeV, provided by the neutron beam
facility installed at the Physikalisch-Technische
Bundesanstalt (PTB), at Braunschweig (Germany).
The measured quantities are:
Ethe distribution of proton recoil response to
monochromatic neutron fields;
Ethe dependence of pulse height and resolution on
energy;
Ethe relation between scintillator outputs for pro-
tons and electrons;
Ethe absolute sensitivity and detection efficiency
of the system.
A dedicated Monte Carlo simulation has suc-
cessfully been used to check the consistency of the
experimental response functions and detection effi-
ciency of our system.
All these results will provide useful input data for
Monte Carlo response calculations with the final
32-cell detector.
2. Experimental set-up
This study covers separate tests of two samples of
BC501A liquid scintillators: one is a 0.3 l cell indus-
trially manufactured by BICRON Corp. (“IM
cell”), the other is a 0.8 l home-made cell (“HM
cell”). In both cases the liquid is encapsulated in
bubble-free cylindrical volumes with a white diffus-
ing coating applied inside. Properties of the
BC501A liquid scintillator are summarized in
Table 1(a) [1]. Dimensions and construction de-
tails of the two cells are reported in Table 1(b).
Both ends of the cells are terminated with Plexiglas
optical windows directly sealed to 2 in PMTs
(EMI-9964KB03).
A dedicated electronic system has been developed
for the PTB calibration test. An improved version of
this electronic system is used to read out data from
the 32-cell detector for n-background measurement
at Gran Sasso. We briefly report here the working
principle of the electronics used during the PTB run.
The first stage of the electronic system is made of
a linear fan-in/fan-out, where the current pulse
286 F. Arneodo et al. /Nucl. Instr. and Meth. in Phys. Res. A 418 (1998) 285—299
Table 1
(a) Properties of the BC501A (BICRON) liquid scintillator (as reported in Refs. [1,2])
Composition Xylene (CH) naphtalene (CH) activators and w.l. shifter
Density 0.874 g cm\
H/C ratio 1.212
Atomic weight 13.2
H atoms per cm4.82;10
C atoms per cm3.98;10
Wavelength of max. emission 425 nm
1q2 of main components (ns) 3.16, 32.3, 270
(b) Construction details of the two cells
IM cell HM cell
Int. length (cm) 15.2 50
Ext. diameter (cm) 8 5
Int. diameter (cm) 5.1 4.4
Material Aluminum #St. Steel (0.15 cm)
Acrylic #Teflon (0.15 cm)
white reflector paint
Sensitive volume (cm) 309 760
Expansion volume (cm)10 30
outputs from the two PMTs are summed together,
followed by a Constant Fraction Discriminator
(CFD) providing trigger signals with precise timing
information.
The subsequent sections of the electronic chain
combine two separate methods of PSD to trigger
and qualify an event as c-ray or neutron with high
identification efficiency.
In the first method (see Fig. 1) the summed signal
from the PMTs viewing the cell is shaped with a RC
filter, i.e. converted into a bipolar pulse. This bi-
polar pulse crosses the baseline at a time uniquely
determined by the nature of the ionizing particle
(independent of amplitude). Hence, measurement of
the time difference between the CFD trigger time
and the zero-crossing time of the bipolar pulse
determines the particle type.
An electronic board containing three stages, the
fan-in/fan-out, the CFD and the zero-crossing PSD
stages, has been developed and built by our collab-
oration for this application [3].
In the second method, the summed signal from
the PMTs is duplicated into two pulses by means of
a linear fan-out, (auxiliary attenuation or amplifica-
tion stages can be used before the pulse splitting).
Moreover, two time gates, separated by a fixed
delay, are generated from the CFD output in coin-
cidence with one of the pulses. Each of these gates
transmits the selected portion of the current pulse
to an integrating 11-bit analog-to-digital converter
(ADC). The electronics in use is schematically re-
ported in Fig. 2.
In our system, the first (300 ns) time gate
covers the whole signal and the second gate (300 ns)
is delayed by 35 ns. This delay is long enough
to exclude the peak of the signal (see Fig. 2).
Therefore, the first integral represents the total
charge, the second (called “delayed charge”)is
the integral of the pulse tail. The relative amount
of charge in the two regions integrated by the
ADCs is characteristic of the type of particle which
caused the scintillation: a higher delayed charge,
relative to the same total charge deposition, will
distinguish a neutron interaction from a gamma
interaction having a lower delayed charge (see de-
tails of Fig. 2).
F. Arneodo et al. /Nucl. Instr. and Meth. in Phys. Res. A 418 (1998) 285—299 287
Fig. 1. Block diagram of the electronics for the “zero-crossing”PSD method, and time relationship between the pulses at various points
of the circuit.
We used commercial NIM standard electronics to
implement this PSD method in our electronic chain.
The two methods are not completely indepen-
dent. We use the prompt zero-crossing information
to generate a neutron flag at trigger level. The
pulse-shape information (total to delayed charge
comparison) for easy handling off-line analysis of
the n-flagged events is provided by the second
method, with effective reduction of unwanted c-ray
events.
The data readout makes use of CAMAC stan-
dard electronics. The data acquisition is performed
by a dedicated LabView application, running on
a Macintosh platform. A trigger veto was imple-
mented during the busy time of the data acquisi-
tion. The busy time of our set-up ranged from 5 to
10 ms, depending on the configuration of the DAQ
set-up. Owing to the high trigger rates recorded
during the event acquisition, a dead-time correc-
tion is required for a precise evaluation of the
detection efficiency (see Section 5).
For each event the total and delayed charge
values are recorded together with a flag that
qualifies the event as c-ray or neutron according to
the zero-crossing time.
3. Data taking
During the calibration test the two cells were
exposed alternately to the PTB neutron beams of
various energies, with the axis perpendicular to the
beam direction at fixed distance (441 cm) from the
source.
The radiation reaching the detector has two
components, i.e. the collimated beam of mono-
chromatic neutrons coming directly from the
production target and the background noise which
is radiation scattered from the materials surround-
ing the detector, the room walls, the detector
mounting, etc. This noise is a mixture of neutrons
with degraded energy and photons produced main-
ly from neutron capture in the environment. As
a general rule, the c-ray production in the target is
negligible.
To separate the two components, at each energy
the detector is successively exposed to the neutron
beam bare and blinded by a polyethylene “shadow
cone”which prevents neutrons produced in the
target from reaching the detector directly. Thus, the
“shadow cone run”allows one to evaluate
the background contamination. At the highest
288 F. Arneodo et al. /Nucl. Instr. and Meth. in Phys. Res. A 418 (1998) 285—299
Fig. 2. Block diagram of the electronics for the “digital charge comparison”method, and time relationship between pulses and
gates. A “Total”vs. “Delayed Charge”plot from off-line analysis is also shown: the upper branch is from neutron interactions and
the lower one is from gamma interactions, both neutrons and gamma’s are emitted from an Am—Be source (37 mBq nominal
activity).
available energy [19 MeV, from T(d, n)He reac-
tion] a correction due to neutrons produced by
spurious reactions in the target support is also
required. This correction is obtained by exposing
the detector (“blank run”) to a neutron beam gener-
ated with a tritium-free target.
The CFD total (n#c) trigger rate was measured
by a dedicated scaler and the relative fraction ( f)of
neutron triggers was successively evaluated by off-
line analysis. Several calibrated detectors from the
PTB facility monitor the beam and allow the value
of the neutron fluence at the detector cell to be
stated. The acquisition rate is lower than the trigger
rate because of the cumulated dead-time of the data
acquisition system. We assume that the acquired
spectrum is unbiased. Table 2 includes the avail-
able neutron energies and the used reactions to-
gether with all the physical quantities required to
obtain the calibration parameters, as described in
the following sections.
F. Arneodo et al. /Nucl. Instr. and Meth. in Phys. Res. A 418 (1998) 285—299 289
Table 2
Neutron energies and calibration parameters. Trigger rate and time refer to the IM cell runs. These values will be used in Sections 4
and 5
Energy Reaction Run Trigger fNo. of Time Neutron
(MeV) type typerate n#c% of n n (s) fluence
(Hz) triggers evts (cm\)
0.565 Li(p, n)Be A 422 75 100k 578.5 1.03$0.05;10
S 152 25 10 k 108.4 1.72$0.09;10
1.2 T(p, n)He A 815 77 100k 458.4 9.19$0.33;10
S 209 30 10 k 81.0 1.64$0.07;10
2.5 T(p, n)He A 350 83 100k 517.2 1.49$0.05;10
S 99 13 10 k 135.0 3.93$0.14;10
5.0 D(d, n)He A 735 80 50 k 237.6 7.54$0.26;10
S 93 23 50 k 697.5 2.21$0.08;10
14.8 T(d, n)He A 2800 73 100 k 430.0 9.68$0.34;10
S 190 29 10 k 51.1 1.12$0.04;10
19.0 T(d, n)He A 700 51 100k 498.6 1.68$0.07;10
S 253 23 10 k 116.5 3.78$0.15;10
B 234 6 50 k 376.0 1.50$0.05;10
A"Acquisition run, S "Shadow-cone run, B "Blank run.
Neutron fluence values are provided by the PTB.
The value corresponding to 19 MeV blank run is estimated by us using the original data from the beam monitors.
Table 3
Data of photon sources used for calibration
Source EA(MeV) E(MeV)
Am 0.060
Na 0.511 0.341
Cs 0.662 0.477
Mn 0.835 0.639
Na 1.275 1.061
K 1.461 1.242
Tl 2.614 2.381
Am/Be 4.430 4.190
4. Calibration procedures
In order to set up a common scale for any detect-
able particle (electrons, protons, alpha’s, etc.), inde-
pendent of the particular ADC conversion factor,
the convention of establishing the pulse-height
scale in terms of “light units”(LU) is adopted [4—6].
The light output for heavy charged particles
¸a(E) is generally expressed in terms of the elec-
tron light ¸(E) output because this quantity has in
practice a linear dependence on the electron energy
(above about 50 keV).
To define this reference electron light output, the
cells were exposed to c-ray sources of various ener-
gies up to 4.4 MeV. As calibration points we used
the photoelectric peak for the Am (EA"60 keV)
and the Compton edges at all the other energies.
Table 3 includes the list of the gamma sources
used for the electron light output definition to-
gether with the corresponding photon energy and
the Compton electron energy.
According to Klein et al. [5,6], we assume
¸"j(E!E), E550 keV (1)
where the energy offset is E"5 keV [4] (to com-
pensate the non-linearity of the light output due
to quenching effects at low electron energies) and
the slope jis set equal to 1 MeV\.
From the gamma source runs we have first to
establish the conversion factor between ADC
counts and light units (the proportional behaviour
is assumed to hold all over the dynamic range of
our ADCs ¸"cN¸"!
). At the same time we want
also to check the assumed linearity of ¸(E) and the
value of the energy offset E.
290 F. Arneodo et al. /Nucl. Instr. and Meth. in Phys. Res. A 418 (1998) 285—299
The light output for protons, corresponding to
neutron energy E, can be converted to the scale in
light units using the gamma calibration results:
¸"cN¸"!
(E), (2)
where cNis the conversion factor from ADC counts
to light output determined by the c-calibration.
4.1. Gamma calibrations
We developed a calibration procedure which ap-
plies on each spectrum measured with the c-ray
sources listed in Table 3. This procedure allows at
the same time
1. to specify the energy calibration point in the
experimental spectrum, taking into account the
finite detector system resolution;
2. to determine the resolution of the detector at the
calibration point.
For this purpose we proceed by steps. First the
spectrum N(¸) (in LU) of Compton electrons
produced in the scintillator system from a given
source (except for the Am source) is generated
by means of the Monte Carlo code GRESP [6]: the
simulation takes into account multiple scattering
and wall effects in a cell with the construction
details properly simulated, according to Table 1.
Resolution smearing is not considered at this stage.
In a second step, the “realistic”Monte Carlo spec-
trum is obtained from the convolution of the unfol-
ded spectrum with the response function of the
detector system:
N+!(H)"R(H,¸)N(¸)d¸, (3)
where all the energies are expressed in LU and His
the discrete ADC counts scale; N(¸)d¸is the differ-
ential number of electrons produced with energy
within d¸about ¸; and R(H,¸)d¸is the differen-
tial probability that an electron of energy within d¸
about ¸leads to a pulse with amplitude H, in ADC
counts.
We assume that for a fixed energy the response
function of the detector is Gaussian, i.e. R(H,¸) can
be written as
R(H,¸)"Aexp!(cH!¸)
2p
*, (4)
where cis the ADC to light conversion factor in
(ADC counts\) for a given calibration source run;
p*"¸o
2(2lg2
with
o"
*¸
¸"detector system resolution FWHM (%),
and Ais an imposed normalization factor between
the Monte Carlo spectrum and the measured one.
The Compton spectrum N(H) defined in this way
depends on two free parameters: the conversion
factor cand the resolution oof the system. These
parameters can be adjusted in order to reproduce
the corresponding spectrum measured in our calib-
ration run. To evaluate the parameters cand owe
used the MINUIT package (CERN Program Li-
brary, D506) to perform a minimization of the
svariable defined as
s"
G
(N(H)!N+!(H))
p
(H), (5)
where N(H) is the experimental spectrum, i.e.
the number of events collected at the ith value of
the ADC range (the averaged pedestal value is
subtracted), N+!(H) is the corresponding Monte
Carlo spectrum and p
(H) is the statistical error
on the counts in the ith bin.
The minimization is extended on a range of ADC
counts (limits of H) selected in order to exclude
bins with low statistical significance at higher am-
plitudes, and bins at lower amplitudes, with signifi-
cant differences, compared to Monte Carlo data,
due to cbackscattering from surrounding materials
not included in the Monte Carlo simulation. As an
example we report in Fig. 3 the unfolded Compton
spectrum from the GRESP code, together with the
experimental spectrum obtained with the HM cell
and the smeared Monte Carlo spectrum, in case of
Cs source.
For the Am source calibration we obtained
the conversion factor cof the calibration scale by
relating the mean value of the measured photo-
electric pulse-height distribution to the energy of
the source, and the resolution oas the full-width at
half-maximum.
F. Arneodo et al. /Nucl. Instr. and Meth. in Phys. Res. A 418 (1998) 285—299 291
Fig. 3. Unfolded spectrum from GRESP (dotted line), experimental spectrum with HM cell (dashed line) and Monte Carlo spectrum
(smoothed line) for Cs source. The position H(ADC counts) of the calibration point (Compton edge ¸in LU) in the experimental
spectrum is also indicated.
The cparameter, calculated for each individual
source run, allows to establish the position (H)of
the calibration point (¸) in the experimental spec-
trum, as indicated in Fig. 3.
The “averaged”value of the calibration factor, cN,
(from ADC to light output, to be used in Eq. (2)),
for both the HM cell and the IM cell, is finally
obtained from a linear fit of the ¸and Hvalues at
the calibration points, relative to the corresponding
set of c-source runs. As an example, in Fig. 4 the
¸and Hvalues, relative to the set of c-sources
taken with the HM cell, are shown as well as the
“averaged”value of the calibration factor.
The light output values ¸at the Compton edge
are plotted versus the Compton energy E!in Fig. 7,
as explained in the following paragraph.
The last result obtained from the gamma calib-
ration procedure is the estimate of the resolution
*¸/¸at the calibration points (i.e. the ovalues
used for folding the simulated response). The
values obtained with the IM cell and the HM
cell are plotted in Fig. 5. These values agree
well with some other resolution measurements [7]
obtained with similar cells and equivalent liquid
scintillator.
According to Ref. [5,6], the resolution *¸of the
detector signal, as a function of ¸, can be given by
the following relation:
D¸
¸"a#b
¸#c
¸
. (6)
It includes three independent contributions:
1. a constant term (a) due to the locus-dependent
light transmission from the scintillator to the
photocathode;
292 F. Arneodo et al. /Nucl. Instr. and Meth. in Phys. Res. A 418 (1998) 285—299
Fig. 4. Linear fit of the ¸and Hvalues from various sources obtained with the HM cell in 0 dB amplification condition.
2. a stochastic term (b) due to the statistical behav-
iour of the light production and attenuation in
the liquid, and of the photon—electron conver-
sion and electron amplification in the PMT;
3. a noise term (c) from the PMT and eventually
from the electronic amplifier.
All these parameters depend on the construction
details of the detector and therefore we expect a dif-
ferent behaviour of the resolution function for the
two cells used at PTB. In our case the contribution
from the noise term is expected to be negligible
compared to the other two terms. Therefore, the
resolution function was determined with a two-
parameter (aand b) fit shown in Fig. 6. The energy
resolution of the home-made cell (HM: a"9.0%,
b"10.8%) turns out to be only slightly worse than
that of the industrially-made cell (IM: a"6.7%,
b"10.2%), owing to the larger size of the active
volume.
4.2. Neutron calibration
Proton recoil pulse-height spectra have been
collected, exposing alternately the two cells (IM
and HM) to the PTB neutron beams at different
energies.
The elaboration of the experimental distribu-
tions, obtained at each of the nominal neutron
energies listed above (Table 2), is carried out ac-
cording to the following sequential steps:
1. The gamma-ray background is eliminated by
selecting only neutron-flagged events and by
F. Arneodo et al. /Nucl. Instr. and Meth. in Phys. Res. A 418 (1998) 285—299 293
Fig. 5. Resolution function and experimental data (cand n) for HM cell (dashed line —a"9.0%, b"10.8%); and IM cell (solid line
—a"6.7%, b"10.2%).
At the highest neutron energy point (19 MeV) the rate distri-
bution from the blank run, after the appropriate normalization,
is also subtracted.
applying a further cut on the delayed-to-total
charge ratio. This cut allows one to discard
residual, unwanted gamma events surviving the
zero-crossing time selection.
2. The pulse-height spectra are corrected for the
acquisition dead-time. [We multiply the counts
of neutron events acquired in each bin by the
ratio of the neutron trigger rate over the total
acquired n-events. The neutron trigger rate is
defined as the total (n#c) trigger rate scaled by
the corresponding fraction of neutron triggers,
f(Table 2):
l(H)"N(H)l>A
)f
N
. (7)
The distribution obtained in this way represents
the distribution of the effective n-trigger rate, l,in
each bin (H)ofthediscreteADCcountsscale.]
3. The neutron background, from scattering with
extraneous materials in the irradiation room, is
eliminated at each energy using information
from the corresponding shadow cone run.
[The rate distribution from the shadow cone
run, obtained according to Eq. (7), is subtracted
after an appropriate normalization, using the
known values of fluence and run time, to reduce
possible differences in the emitted n-fluxes be-
tween the acquisition run and the shadow-cone
run:
l"l!l1
/*t
1/*t1
, (8)
294 F. Arneodo et al. /Nucl. Instr. and Meth. in Phys. Res. A 418 (1998) 285—299
Fig. 6. Proton recoil response to mono-energetic neutron beams at 5.0 (solid histogram), 14.8 (dashed histogram) and 19.0 MeV (dotted
histogram), for the 0.3 l IM cell. The NRESP Monte Carlo distribution at 14.8 MeV, smeared by the corresponding detector resolution
(*¸/¸"7.8%), is also shown (solid line).
where l is the net n-trigger rate in each bin,
the fluence and *tthe run time from Table 2.
A and S refer to acquisition and shadow run
respectively.]
4. The distributions are finally normalized in such
a way that their area is equal to the net, effective
neutron trigger counts per unit of fluence.
[The total net counts in each bin (i.e. the effective
number of monochromatic neutron interactions
per bin at a given energy) per unit of fluence is:
N"l
)*t
. (9)
The N distribution represents the effective re-
sponse function to monochromatic neutrons.]
The resulting response distributions from the IM
cell, corresponding to 5, 14.8 and 19 MeV of neu-
tron energy, are shown in Fig. 6. The vertical scale
represents the number of recoils per energy bin, per
unit of neutron fluence. Recoil spectra at the lower,
available neutron energy of 0.565 MeV were slight-
ly suppressed on account of the energy threshold
(E
). The analysis of this point is affected by larger
systematic effects.
As expected, the response functions do not show
a vertical, sharp edge at the end point, owing to the
finite resolution of the detector.
In the analysis of the experimental distributions,
we have applied, at each energy point, a procedure
similar to that used for the ccalibrations (Sec-
tion 4.1): we used as unfolded proton spectrum
F. Arneodo et al. /Nucl. Instr. and Meth. in Phys. Res. A 418 (1998) 285—299 295
The NRESP code (version 7) performs Monte Carlo calcu-
lations of the response function from fast neutron interactions in
liquid scintillator cells. The neutron source and the detector
geometry can be properly defined by the user. The code makes
use of an extensive set of cross-section data. It includes neutron
scattering effects in the detector housing and secondary particle
production.
The NEFF Monte Carlo code is part of the NRESP pro-
gram, dedicated to the efficiency calculations.
a distribution generated by means of a dedicated
Monte Carlo code NRESP[2], convoluted with
a Gaussian response function of the detector. It is
important to stress that the light output function
for protons is not linear with respect to the energy.
The procedure used to determine the position
¸"!
(E) of the upper edge and the resolution can be
then applied only locally in a neighbourhood of the
upper edge of the response function. As an example,
the Monte Carlo distribution at 14.8 MeV is super-
imposed in Fig. 6 on the experimental spectrum,
showing a satisfactory agreement.
This procedure allows one to determine the fol-
lowing experimental quantities, objectives of the
PTB test:
(a) The light output function, ¸(E) for protons
(Fig. 7), from neutron interactions, determined by
detecting the position of the upper edge of the
response distribution.
(b) The energy resolution, *¸/¸(%), as a function
of the energy deposited, estimated by fitting
the width of the folding Gaussian on simulated
data to reproduce the experimental response func-
tion. The energy resolution values obtained from
the neutron runs with the IM cell are reported on
Fig. 5. These points have been used, together with
those from the gamma calibration runs, to deter-
mine the resolution function, also reported on the
plot.
In Fig. 7 the dependence of light output for
protons is displayed vs energy, for both the IM
and HM cells. For comparison the equivalent
functions for electrons, from gamma calibrations,
and a-rays are also shown. Available experi-
mental light output data and parameterization
from Refs. [8—10] are reported in Fig. 7. More
recent measurements from Ref. [11] have been
checked to be also in very good agreement with our
results.
5. Neutron sensitivity
The neutron sensitivity, S, expressed as total
counts/( fluence at the counter), for a collimated neu-
tron beam at a given neutron energy, can be easily
obtained by integrating the corresponding re-
sponse function, previously defined in Eq. (7), over
the energy range from E
and E
"E. The bias
level E
is calculated from the corresponding thre-
shold expressed in light units, using the calibration
curve reported in Fig. 7.
Values of the sensitivity S, and relative neutron
energy and bias level, for the IM cell are reported in
Table 4. The error on the sensitivity has been
evaluated assuming a 5% relative error on the
determination of the trigger rate and on the neu-
tron fraction, and a 3% error on the exposure time.
The error on the neutron fluence at different ener-
gies has been provided by PTB and is reported
in Table 2.
The sensitivity is strongly affected by the bias
setting: a high value of E
, compared to the inci-
dent neutron energy, cuts a large fraction of proton
recoils, thus limiting the sensitivity to a low value,
as in the case of our measurement at 0.565 and
2.5 MeV, where more than 50% of the proton sig-
nal is cut away.
The sensitivity expected with the IM cell has
been calculated by means of the NEFF Monte
Carlo code[2] at the various experimental condi-
tions (bias settings as reported in Table 4).
The results are shown in Fig. 8 together with our
experimental data. The agreement between our re-
sults and the corresponding Monte Carlo values
expected is satisfactory.
In addition to the sensitivity, another quantity
may be used, the detection efficiency e", i.e. the
probability that a neutron which enters the de-
tector will be detected. It can be obtained from the
sensitivity, dividing its value by the area, A", of the
detector perpendicular to the incident neutron
beam direction (A""77.5 cmfor the IM cell,
e"values are reported in Table 4).
296 F. Arneodo et al. /Nucl. Instr. and Meth. in Phys. Res. A 418 (1998) 285—299
Fig. 7. Light output data for protons, electrons and alpha’s with the IM cell and HM cell. The ameasurement was performed with
aU source (E?"4.2 MeV), deposited on a small metallic layer, and dipped into the active volume of the HM cell after the PTB test.
Dashed line: expected electron light output ¸. Solid line and open squares: proton light output from Refs. [6,7]. Solid squares: alpha
light output from Ref. [7].
Table 4
Absolute sensitivity, S, and detection efficiency obtained with
IM cell. Corresponding neutron energy and bias level are also
reported
E(MeV) E
(MeV) S(cm)e"(%)
0.565 0.25 15.4$1.4 19.9
1.2 0.25 28.2$2.4 36.5
2.5 1.4 9.6$0.8 12.4
5.0 1.4 17.8$1.6 23.1
14.8 2.0 8.8$0.7 11.4
19.0 2.0 8.5$0.9 11.0
Assuming a cylindrical shape of the liquid scintil-
lator cell, the efficiency depends only on the dimen-
sion parallel to the beam, in our case the diameter,
independently of the transverse dimension (A").
Therefore, our result applies as well as for detectors
with equivalent radius but different transverse
areas perpendicular to the beam.
The IM cell has the same (internal) radius but
smaller length than the cells of the spectrometer for
the Gran Sasso neutron background measurement.
This allows us to state that the sensitivity, here
evaluated with collimated beams, may provide suit-
able information for the deconvolution of a recoil
spectrum measured in an isotropic neutron field, as
expected in the case of a measurement at the Gran
Sasso underground laboratory. In that case the
sensitivity has to be scaled by a factor p/4, as
demonstrated in the appendix.
The expected fission spectrum of neutrons from
natural radioactivity at Gran Sasso ranges up to
about 15 MeV. Sensitivity at various (collimated
F. Arneodo et al. /Nucl. Instr. and Meth. in Phys. Res. A 418 (1998) 285—299 297
Fig. 8. Neutron detection sensitivity (counts per unit fluence) versus neutron energy with IM cell. Experimental data (solid points) and
Monte Carlo output (histogram) corresponding to different bias settings.
Table 5
Absolute sensitivity, S(cm), for various (collimated beam) neutron energies (MeV) and different bias levels (MeV)
E
E(MeV)
(MeV)
1.0 2.5 5.0 7.5 10.0 15.0 20.0
0.65 23.7 24.9 20.0 16.6 14.5 13.8 12.3
0.80 15.2 23.4 19.3 16.0 14.0 13.4 12.1
1.00 8.6 21.4 18.5 15.8 13.4 12.7 11.9
beam) neutron energies (from 1 up to 20 MeV),
with different bias settings (E
"0.65, 0.80 and
1.0 MeV, corresponding to 0.1, 0.15 and 0.2 ex-
pressed in light units), was calculated by NEFF
Monte Carlo simulation and is reported in Table 5.
These data should provide useful information for
unfolding purposes.
6. Conclusion
Two cells, industrial and home-made, filled with
a hydrogenated liquid scintillator, BC501A, have
been calibrated with monochromatic neutron
beams and gamma sources at the PTB Accelerator
Facility.
298 F. Arneodo et al. /Nucl. Instr. and Meth. in Phys. Res. A 418 (1998) 285—299
The performances and the reliability of both the
HM cell, compared to the industrially-made cell,
and of the electronic system have been successfully
tested during the PTB run. In particular, the light
output response of the scintillating medium to vari-
ous particles (electrons, proton and alpha’s) has
been measured. It was found to be in very good
agreement with other published experimental data.
The energy resolution of our set-up has been
quoted; it seems compatible for the underground
neutron measurement. Finally, the neutron sensi-
tivity and the detection efficiency have also been
measured and compared with the results of our
dedicated Monte Carlo simulation, showing a satis-
factory agreement.
Based on these results, a detector consisting of an
array of 32 home-made cells, with an improved
version of the electronic layout, has been developed
and built by the ICARUS Collaboration. All the
information collected during the PTB test and re-
ported here will provide us with the necessary tools
to perform flux and energy spectra measurement of
neutrons from natural radioacitivity in Hall C, at
the underground Gran Sasso Laboratory.
Acknowledgements
Special thanks go to H. Klein, for fruitful sugges-
tions on many aspects of the liquid scintillator
technique and for providing us with the GRESP
and NRESP Monte Carlo codes. We would like to
thank D. Guldbakke and collaborators for their
precious help during the test period at PTB Labor-
atory. The technical support of W. Galli and F. Del
Grande, E. Tatananni and B. Romualdi, A. Fal-
giani and M. Giusti, O. Barnaba` and G. Musitelli is
also acknowledged.
Appendix A
In case of an isotropic neutron field the response
function, and consequently the sensitivity, is differ-
ent if compared to the response function obtained
with a collimated neutron beam. To get a trans-
formation from one case to the other, it is necessary
to know the flux in an isotropic field, which would
produce the same rate of neutrons crossing the
counter as that obtained in the directional field. In
our case, i.e. considering a cylindrical shaped de-
tector and disregarding the neutrons which enter
the cylinder from the two bases, if U and U are
the fluxes in an isotropic and in a collimated field,
the rate of neutrons which cross the counter in the
two situations are, respectively [12],
U
nR¸
2and U 2R¸,
where Rand ¸are the radius and the length of the
cylinder. They must be equal, hence
U"U
4
p.
The sensitivity being inversely proportional to the
flux, the sensitivity in an isotropic field is then
S"S
p
4.
In this way the sensitivity measured at PTB can be
easily transformed into the expected sensitivity
of our detector when used at the Gran Sasso
Laboratory for the underground neutron flux
measurement.
References
[1] BICRON Corp., BC-501A Liquid Scintillator Data
Sheets.
[2] G. Dietze, H. Klein, NRESP Monte Carlo, PTB-ND-22,
1982.
[3] O. Barnaba` et al., Nucl. Instr. and Meth. A 410 (1998) 220.
[4] G. Dietze, IEEE Trans. Nucl. Sci. NS-26 (1979) 398.
[5] G. Dietze, H. Klein, Nucl. Instr. and Meth. 193 (1982) 549.
[6] M. Tichy, H. Klein, J. Pulpan, PTB-7.2-92-1, 1992.
[7] H.H. Knox, T.G. Miller, Nucl. Instr. and Meth. 101 (1971)
519.
[8] R.A. Cecil, B.D. Anderson, R. Madey, Nucl. Instr. and
Meth. 161 (1979) 439.
[9] M. Anghinolfi et al., Nucl. Instr. and Meth. 165 (1979) 217.
[10] V.V. Verbinski et al., Nucl. Instr. and Meth. 65 (1968) 8.
[11] J.H. Lee, C.S. Lee, Nucl. Instr. and Meth. A 402 (1998) 147.
[12] K.H. Beckurst, K. Wirtz, Neutron Physics, Springer,
Berlin, 1964, p. 248.
F. Arneodo et al. /Nucl. Instr. and Meth. in Phys. Res. A 418 (1998) 285—299 299
