Nuclear Instruments and Methods in Physics Research A 394 (1997) 332-340
NUCLEAR
INSTRUMENTS
8 METHODS
IN PHYSICS
RESEARCH
Section A
ELSEVIER
Detection of relativistic neutrons by BaF2 scintillators
V. Wagnera**, A. Kugler”, M. Pa&r”, M. &unberaa, A. Taranenkoa,
S. HlavSCb, R. Lorencz’, R. WohlgemuthC, R.S. Simond
=Nuclear Physics Institute of‘ ASCR. CZ-250 68 pe?, Czech Republic
bInstitute of’ Physics of SAS, SK-842 28 Bratislava, Slovak Republic
CLudwig-Maximilians-lJniversitiit Miinchen, D-85748 Garching, Germany
dGesellschaft ftir Schwerionenforschung, D-64220 Darmstadt, Germany
Received
19 February
1997
Abstract
Neutrons and photons with energies from 100 MeV to 1.3 GeV are registered in a cluster of seven hexagonal BaF2
scintillator modules. The neutron-induced hadronic shower is studied as a function of the incident neutron energy and is
compared with the photon-induced electromagnetic shower. From the neutron flux and spectral distribution at the position
of the BaF2 detector, which were measured with a calibrated liquid scintillation detector, the absolute efficiency for neutron
detection is determined as a function of the incident energy. A simple exponential relation is used to extrapolate to zero
electronic threshold. The results are compared with GEANT3 simulations.
PACS: 29.40.M; 25.75.-q; 25.40.f
Keywords:
Relativistic neutron detection;
Barium fluoride detectors;
1. Introduction
Barium fluoride detectors are now commonly used to detect both high-energy photons and light charged particles.
BaF2 has good time and energy resolution as well as high
efficiency. Recently, it was shown that BaF2 scintillator also
is an efficient detector for neutrons in the range of several
10 MeV, having a time resolution comparable to that of
the classic NE213 liquid scintillator [l-3]. Here we extend
these investigations to relativistic neutrons with energies up
to 1300 MeV. The data were collected in Bi+Pb collisions
of 1 GeV/u at the heavy-ion synchrotron at GSI Darmstadt.
2. Measurements and analysis
2.1. Experimental arrangement
The experimental setup consisted of a closely packed
cluster of seven hexagonal BaF2 modules and a cylindri-
* Corresponding author. Tel.: +420 2 6617 3170; fax: +420 2
6857 7003; e-mail: [email protected].
016%9002/97/$17.00
Copyright
PII SO168-9002(97)00674-8
Hadronic
shower; Electromagnetic
shower
cal NE213 liquid scintillation (LQ) detector. The BaFz detector was positioned above and the LQ detector was positioned below the horizontal plane through the beam axis. The
polar angles defined for the centers of the front face of the
two detectors were identical. Hence, the LQ detector allowed
us to determine the neutron flux and its spectral composition
at the position of the BaF2 detector. The distance from the
target to the front face of the detectors was d = 4.3 m. The
measurements were performed for the polar angles 29= 23”,
40”, 60” and 90”.
Each individual BaF2 module consisted of a hexagonal
BaF2 crystal (inscribed diameter 59 mm, length 250 mm)
coupled to a Hamamatsu R2059-01 fast phototube. Details
can be found in Refs. [&6] describing the two-arm photon spectrometer TAPS. The reference LQ detector was of
cylindrical shape (diameter 120 mm, length 40 mm). This
detector was read out via a XP2041 phototube. Thin plastic scintillators were placed in front of both the detectors to
identify charged particles. The thickness of these veto detectors were 9 and 10 mm for the LQ detector and the BaF2
detector, respectively.
Events involving the BaFz or the LQ detector were
classified as charged if the energy-loss amplitude in the
corresponding veto detector was above the threshold for
0 1997 Elsevier Science B.V. All rights reserved
V. Wagner et al. I Nucl. Instr. and Meth. in
minimum-ionizing
particles and if the time difference between the two detectors was 5 2 ns, with the timing of
the BaFz cluster being defined by the module with the
highest amplitude. The minimum-ionization
threshold was
determined during the off-line analysis using data from
high-energy pions produced in the BifPb reactions as well
as data from cosmic-ray muons. This analysis established
that the actual electronic threshold for the veto detectors was
safely below the energy deposited by minimum-ionizing
particles.
333
Phys. Res. A 394 (I 997) 332-340
m
2
t
2 lo4
1
”
1.OGeV
1
10000 7,
0.1 GeV
2.2. Data acquisition
2p = 23 deg
In the measurements, a beam-foil detector mounted upstream from the target provided the time reference. Energy
and time signals were recorded for the individual BaF2 modules, for the LQ detector and for both the veto detectors. In
order to exploit the pulse-shape capability of BaF2 scintillator, two integration gates of 50 ns and 2 us. respectively,
were applied to the BaF2 energy signals. For each event also
the multiplicity of charged particles detected in a segmented
plastic-scintillator
forward-wall covering polar angles 7”30” was recorded. see Ref. [7].
2.3. Particle kinetic energies using time-of-jight
After correction for walk and crosstalk effects in the electronic system of the BaF2 cluster detector, we achieved a
time resolution for photons of At = 353 ps FWHM, see the
prompt photon peak in Fig. 1. The time resolution of the
LQ detector was 450 ps FWHM. Given the measured timeof-flight t. one can calculate the kinetic energy per nucleon
&n of the detected nucleons and nuclear fragments using
the nominal target-detector distance d.
For the timing of protons and nuclear fragments the
absolute time calibration based on the prompt photon peak
is immediately applicable. Because of the short radiation
length of BaF2 (X0 = 2.05 cm) the finite length of the detector modules plays a similar role for the timing of charged
particles as it does for the timing of photons. In both cases,
light production essentially starts at the front end of the detector. Furthermore, the walk correction applied to the data
approximately accounts for the fact that the electromagnetic
shower or high-energy protons travel faster than the scintillation light. Thus, the kinetic energy of charged particles
is directly provided by the calculated energy Ecaic as obtained from the time-of-flight information, with the energy
resolution being determined by the time resolution of the
detector.
The situation for neutrons is different, because there is
an additional contribution to the uncertainty in the timeof-flight information apart from the intrinsic detector resolution. This neutron-specific
contribution is due to the
uncertainty Al which corresponds to the variation of the
actual interaction point of the neutron along the length I of
the detector. In the BaF2 modules with I= 25 cm, the com-
20
30
50
40
Time
[nsl
Fig. 1. The experimental time-of-flight spectrum for neutral hits
measured with the BaF2 cluster detector. The sharp peak which has
a resolution of 353 ps FWHM is due to prompt photons while the
broad distribution of later hits is due to neutrons. The absolute time
scale corresponds to arrival at the front face of the detector. An
additional transit time of 1.25 ns for the scintillation light through
the BaF2 module is implied for the walk-corrected prompt photon
signals. The flight times t of neutrons with incident kinetic energies
&n=0.1,0.2,0.3,0.5,
1.0,1.5 and 2.0GeV are indicated.
bined uncertainty in time and position leads to a variation
of rr(EC,~C)/ECa~C
between 2% and 15% for neutron energies
between 100 MeV and 1.5 GeV. For neutron kinetic energies around 350 MeV the influence of Al is small. These
energies correspond to velocities p z l/n, where n = 1.5 is
the refractive index of BaF2. The poor correlation between
incident kinetic energy and observed pulse height also
excludes a systematic effect of the walk correction.
2.4. Energy calibration of the BaF2 signal
As can be seen in Fig. 2, the detection of protons and
heavier fragments leads to pronounced peaks in the BaF2
amplitude spectrum. The proton peaks were used to establish the relation between the output signal for each module
and the calculated kinetic energy Ecalcderived from the BaFz
time-of-flight information and, hence, to define the protonequivalent energy calibration (MeV,,). For this calibration,
the energy loss of the protons along their flight path in air
and in the veto scintillator as well as the resulting retardation were taken into account. The maximum energy which a
proton can deposit in a BaF2 scintillator of 25 cm thickness
by ionization is about 380MeV. Protons with higher-energy
punch through the crystal. Their ionization loss decreases
with increasing incident energy; see, for instance, the spectrum for protons and nuclear fragments corresponding to a
334
V. Wagner et al. I Nucl. Instr. and Meth. in Phys. Rex A 394 (I 997) 332-340
Table 1
Calculated neutron efficiency of the NE2 13 reference detector and
measured neutron efficiency of the investigated BaF2 cluster detector (see Eq. (1)). The reference detector has the shape of a cylinder
with 120 mm diameter and 40 mm length, while the BaF2 cluster
detector is a closely packed array of 7 hexagonal modules, each
having a diameter of 59mm and a length of 250mm. The incident neutron flux is perpendicular to the front face of the detectors.
The efficiency of the reference detector is calculated for the actual
electronic threshold of 4 MeV,,, used in the present experiment.
The efficiency of the BaF2 detector on the other hand is given by
the two parameters sc and i, which provide the BaF2 efficiency at
zero electronic threshold and describe the exponential dependence
of the efficiency on the threshold value (see Eq. (2)). The errors
quoted are statistical only and do not include the 20% uncertainty
estimated for the calculation of the efficiency of the reference detector
I
0
100
I I I, I I1 I
200
300
I,
,,,,I
400
500
L [MeV,,l
Fig. 2. Pulse height spectra measured with the BaF2 cluster detector
for charged particles with EFatc = 200,300,400
and 800MeV/u
(bin width was +5%). Due to the additional coincidence with
the veto detector, random background contributions are negligible.
The broad structure of the proton peak at 400 MeV/u is due to the
rapid decrease of the energy loss above punch through. The energy
calibration corresponds to proton-equivalent energy.
kinetic energy of 800 MeV/u in Fig. 2. In the measurements,
a common energy range of amplitudes from 4.5-600 MeV,,
was covered by all BaFz modules.
The electron-equivalent
energy calibration (Me&) pertaining to the electromagnetic
shower is based on two
calibration points from cosmic-ray muons. Lateral or longitudinal crossing of the BaF2 crystals by muons was
established with external tracking detectors. The observed
amplitudes provide two points on the electron-equivalent
scale at 38 Me& and 168 Me& respectively [8,9].
3. Monte-Carlo simulations
3.1. Monte-Carlo
description of the NE213 scintillator
The neutron detection in the liquid scintillation material
NE2 13 is well described by specialized Monte-Carlo codes
up to a neutron energy of 700 MeV, see Refs. [lO,ll]. We
have extended this range to the maximum neutron energies
studied in the present investigation.
Crucial for this extension to high energies is the knowledge of the relevant neutron cross sections for reactions with
the protons and carbon nuclei of the LQ scintillator material. The elastic and inelastic np scattering is well known
experimentally. But for the n+C reaction at neutron energies above several 100 MeV, where this reaction starts
to dominate the detection efficiency, the only experimental
Ekrn (MeV)
Neutron energy
NE213
Efficiency
BaF2
Efficiency
100
150
200
250
300
350
400
450
500
600
700
800
900
1000
1250
0.041
0.037
0.034
0.033
0.034
0.033
0.032
0.035
0.036
0.037
0.038
0.039
0.037
0.036
0.035
0.28
0.26
0.24
0.24
0.24
0.24
0.24
0.25
0.26
0.28
0.29
0.30
0.29
0.29
0.30
(1)
(1)
(1)
(I )
(I )
(I )
(I )
(1)
( 1)
(1)
(1)
(I)
(I)
(I)
(3)
i. (MeV-‘)
sc
0.0273 (IO)
0.0156 (4)
0.0114 (3)
0.0093 (2)
0.0079 (2)
0.0066 ( I )
0.0059 (1)
0.0058 (2)
information
available
is the total n+C cross section. We
therefore used the intranuclear cascade model [ 121 to calculate the charged-particle producing inelastic cross sections
and attributed the difference between the known experimental total cross section and the calculated inelastic cross sections to elastic scattering. At high neutron energies, elastic
scattering contributes to the detection efficiency because the
recoiling C nuclei also produce detectable light signals.
The resulting efficiency of the reference LQ detector is
given in Table 1 for incident neutron energies between 100
and 1250 MeV. The efficiency of the NE213 reference detector is calculated for the experimental threshold of 4 MeV, the
systematic uncertainty being ~20%. The known response
of the LQ detector was used to determine the neutron flux
and its spectral composition at the BaF2 detector position.
3.2. GEANT
simulation of the BaF2 cluster detector
The Monte-Carlo code GEANT3 [ 131 with the hadronic
shower generator FLUKA was used to simulate the neutron
1/. Wagner et al. INucl.
335
Instr. and Meth. in Ph,vs. Rex A 394 (1997) 332-340
and the photon response of the BaF2 detector. The
geometry of the detector comprises the active volumes of
the seven individual BaF2 modules and the passive 1 mm
gaps between adjacent crystals which are occupied by the
teflon reflector and the light seal. The veto detector is simulated by a 10 mm thick sheet of plastic scintillator which
has the shape of a large hexagon that covers the front face
of the BaFz cluster. Secondary particles due to the hadronic
and electromagnetic showers were tracked to a kinetic energy of 100 keV. We convoluted the energy deposit in BaF2
with a Gaussian of LIE/E = 3% x (E/GeV))“4.
see Ref.
[6]. We also accounted for the quenching of the scintillation
light in the BaF2 crystals known to occur for protons and
nuclear fragments, see Ref. [ 141.
In the experiment both the target and the detector were
positioned 2 m above the concrete floor of the target area.
Scattered neutrons therefore have a much longer flight path
than the direct target-detector
distance of d = 4.3 m. We
have extended the Monte-Carlo simulations to include the
floor of the target area. The calculations show that the contribution due to scattered neutrons and misidentified delayed
photons is 25%, 7% and below 1% for time-of-flight values
corresponding to direct neutrons with kinetic energies of 80,
100 and above 150 MeV. respectively.
3.3. Unfolding the neutron kinetic energy distribution as
calculated front time-of--flight
An important step in the analysis is to establish the relation
between the observed time-of-flight and the incident kinetic
energy of the neutrons. From the simulation the calculated
time for arrival of the light pulse at the photocathode is given
by t,,th = tint+ ttrans,
where t,,t is the time of the first neutron
interaction in BaF2 obtained from GEANT3, while ttlansis
the transit time of the scintillation light travelling from the
interaction point to the phototube. For consistency with the
measured time-of-flight t we fold tcaa with the experimental
time resolution and use an absolute time calibration based
on arrival at the front face of the detector (1 = t&h - 1.25ns).
Following the procedure used in the data analysis we then
determine Ecaic
A sharp primary neutron energy Ekin results in a distribution of kinetic energies EC,,, for the registered neutrons.
Equivalently, a given window in Ecalcreceives contributions
from a distribution of primary energies Ektn which extends
both below and above the window limits of Ecalc.This effect
is demonstrated in Fig. 3. The bin sizes in Ecalcare 50 MeV
for E,,i, up to 800 MeV and 150 MeV for Ecaic higher than
800 MeV, respectively, and we assume a flat energy spectrum. The corresponding distributions in the primary neutron kinetic energy Ekinare shown by the dashed histograms
in Fig. 3. The width of the primary distributions strongly increases with increasing neutron energy and the distributions
become asymmetric extending towards higher values Of Ekin.
Introduction of a more realistic neutron source which represents the projectile participants observed at 19= 23” leads to
800
MeV
1200
0
500
1000
MeV
2000
1500
E,,, [MeVI
Fig. 3. Simulated distributions of the primary neutron kinetic energy
,!?kin for different values of the calculated kinetic energy Ecatc as
determined from the time-of-flight
information provided by the
BaF2 cluster detector. The target-detector
distance is 430 cm and
an absolute time calibration based on arrival at the front face of the
detector is used. The time resolution of the BaF2 cluster detector
is 353~s FWHM, equal to the value achieved in the experiment.
The regions in EC,), are Ecatc = 200 f 25, 400 + 25, 800 zt 25,
1200 & 75 and 1600 2~ 75 MeV. respectively. Solid and dashed
histograms correspond to different assumptions for the incident
neutron spectral distributions (see text).
more symmetrical shapes as shown by the solid histograms
in Fig. 3, simply because in the upper bins of Ecalc the
intensity is now concentrated near the lower edge of the bin.
Using a time calibration based on walk-corrected prompt
photon signals has the consequence that for neutrons with
velocities b > 0.67 which interact deep inside the module, the calculated energies Ecalc are systematically higher
than the true incident energies Ekin; see for instance the distributions corresponding to the 1600 MeV bin in Fig. 3.
The horizontal arrows given in the figure indicate the centroid shift of the reconstructed Ekin distributions as obtained
for the spectral shape expected for projectile participants.
Due to the uncertainty in the interaction point along the extended detector, a correction of this effect is not available on
an event-by-event basis. Instead, the distribution of calculated kinetic energies corresponding to a narrow time window is unfolded to give the average incident neutron kinetic
energy Ek,“.
4. Neutron detection in the BaFz cluster detector
In the following, we investigate the neutron response of
the BaFz cluster detector as a function ofthe incident neutron
336
V. Wagner et al. I Nucl. Instr. and Meth. in Phys. Res. A 394 (1997) 332-340
kinetic energy. The primary experimental variable, however,
is not the kinetic energy itself but rather the neutron timeof-flight. In the analysis, narrow time windows are selected
which correspond to some range of calculated energies. As
discussed in Section 3.3, the distribution of these calculated energies then is unfolded to give the average incident
neutron kinetic energy &n relevant for the time bin under
study.
The first-level approach is to consider the cluster of seven
BaF2 modules as one single large modularized BaF2 detector. By the pulse height of the cluster detector we mean the
sum of the gain-matched amplitudes of all seven modules.
The time information is derived from the module with the
highest individual amplitude. In order to contribute to the
amplitude, the additional modules have to fall into a narrow
time window of &3 ns relative to the cluster time defined
by the module with the maximum signal. For an electronic
threshold of 4.5 MeV,, this time condition leads to the acceptance of ~98% of the events with more than one active
module and keeps the contamination due to random background contributions small.
In a second step we then investigate the response of the
cluster detector on a more detailed level and study the development of the neutron or photon-induced shower. Within the
geometric limitations of the present setup such an analysis
is restricted to events where the maximal amplitude occurs
in the central module of the cluster.
L lMeV,,l
Fig. 4. Pulse height spectra measured with the BaF2 cluster detector for neutrons with incident kinetic energy Ekin = 200,300,400
and 800MeV (the bin width in the underlying calculated energy
E talc was *%I). The contributions corresponding to the random
background in the time-of-flight spectrum were determined from
equally wide time bins at t > 50ns (see Fig. I ) and were subtracted from the raw data. The energy calibration corresponds to
proton-equivalent
energy.
4. I. The pulse height response for neutrons
Protons and nuclear fragments of fixed energy give
rise to pronounced peaks in the response function which
reflect the well-defined energy-dependent
ionization loss of
charged particles. A totally different situation is observed
for neutrons (see Fig. 4). The general feature of the amplitude spectra is their exponential shape without pronounced
peaks. Obviously, the measured pulse height is only poorly
related to the original incident neutron kinetic energy. This
directly shows the random character of the detection process. There also is no sudden drop in intensity above the
proton punch-through energy. Obviously, the interaction of
high-energy neutrons with the Ba and F nuclei of the detector material involves complex reaction channels which
lead to larger energy deposition than the simple (r&p)channel.
The GEANT simulations show that the neutron-induced
reactions in the BaF2 material which lead to substantial energy deposits frequently involve protons in the exit channel.
The resulting pulse height spectra observed for the BaF2
cluster detector are therefore calibrated in terms of protonequivalent energy.
4.2. The absolute neutron detection ejiciency
The neutron flux and its spectral composition at the position of the BaF2 cluster detector were measured with the
NE2 13 reference detector. Given this information the absolute effi&nCy
&BaF1(EkIn)
for the detection of neutrons by
the BaF2 cluster detector can be determined as a function of
the incident neutron energy Ekin according to the relation
(1)
are count rates and solid
NB~F:,
NLQ,
QB~F?, &Q
angles for the BaFz cluster detector and the LQ detector,
respectively. The detection efficiency &LQ(.!!&)
of the reference LQ detector was computed by a Monte-Carlo code.
The overall error of ELMis estimated to be below 20% and it
is not included in the errors of the EB~F~values discussed in
the present study. Due to the nearly exponential shapes of
the amplitude spectra (see Fig. 4) the efficiency strongly depends on the value Lthr of the electronic threshold employed
in the measurement.
The absolute efficiencies of the BaF2 cluster detector
EB~F~(&,) are plotted in Fig. 5 as a function of the neutron
kinetic energy &i,, for different values of Lthr. The detection
efficiency increases with increasing energy of the neutrons
and reaches a plateau above 750 MeV.
The GEANT3 simulations give EB~F~values which are
consistently higher than the data. However, if the simulated
values are scaled down by a factor of x0.75, they reproduce
the shape of the data quite well up to 750 MeV.
where
331
V. Wagner et al. I Nucl. Instr. and Meth. in Phys. Res. A 394 (1997) 332-340
x
2
.?
.u
E
w
0”
0.4
.-5
L,=OMeV,
L,=SMeV,
.0
+
G
-1
10
70
Lw,=90MeV,
-2
10
-:
IO
0
0
E,,, [MeVI
lOOh4eV
0
-
v
-
150kkV
*
-
300
t&Y
A
-
500
t&V
n
-
1200MeV
-I
20
5
40
60
80
IO
100
15
20
25
30
LmR [MeV,l
Fig. 5. The neutron efficiencies of the
various values of the electronic threshold
(only statistical errors). The extrapolation
GEANT3 results multiplied by 0.75 are
BaFz cluster detector for
Lb as a function of ,!&#a
to Lmr = 0 is also shown.
given by full lines.
Fig. 6. The neutron efficiencies of the BaF2 cluster detector for
various incident neutron kinetic energies ,!?kia as a function of &,,.
The straight lines show the least-squares fits to the data.
200 MeV
L bfIeV,,l
L tMeV,l
1?
C
/
0’
0
0
200
400
L [MeV,l
a
t
EKIN= 800
MeV
lo3
0
200
400
L [MeV,,l
Fig. 7. Pulse height spectra as observed in the central module (a), in the whole BaF2 cluster detector (b) and in the shielded central
module (c), respectively (.& = 4.5 MeVpe). For the latter case, the absence of a signal in the surrounding modules was required. All events
correspond to central hits selected by the condition that the maximum signal occurs in the central module. In this respect the spectra of
case (b) differ from the unconditional spectra of Fig. 4.
V. Wagner et al. I Nucl. Instr. and Meth. in Phys. Rex A 394 (1997) 332-340
338
The nearly exponential pulse height response for a given
incident neutron energy is because the detection efficiency
strongly depends on the electronic threshold Lthr, see Fig. 4.
To examine this dependence more closely, we varied the
threshold values for the BaFz cluster detector in the off-line
analysis from 4.5 up to 90 MeV,,. The resulting efficiencies
follow quite well a simple exponential dependence on ,&,
see Fig. 6. This general behavior is independent of the particular value of the incident neutron kinetic energy, while
the slope does depend on ,!?kin.The straight lines shown in
Fig. 6 correspond to least-squares fits using the relation
dEkin)=
&O(Ekin)eXP(-~(Ekm)
X Lthr),
(2)
where EOis the efficiency in the limit of zero threshold and
L is the slope parameter. The applicability of a parametrization as suggested by Eq. (2) agrees with the observations reported for lower neutron energies in Ref. [2]. Both EOand i
are functions of&n. The numerical results for the detector
under study are given in Table 1.
At neutron energies around lOOMeV, the efficiency of
the BaFz cluster detector is higher than the efficiency of the
NE213 reference detector by a factor given approximately
by the ratio of the lengths 1 of both detectors, provided the
same electronic thresholds of ,& = 4 MeV,, are applied.
Hence, in agreement with the study of Ref. [3], we find that
the efficiency of a BaF2 detector is comparable with that of
a NE2 13 detector of the same volume for neutron energies
around Ekln = 1OOMeV. The ratio (EB,F,/~B,F~)/(&LQ/~LQ),
however, starts to deviate from unity with increasing energy of the incident neutrons and reaches a value of 1.4
at 1.3 GeV.
4.3. Comparison
of neutron and photon-induced
showers
We assume that both neutrons and photons deposit most
of their energy loss in the module encountered first. For an
analysis of the shower properties as a function of the incident
energy, we therefore restrict ourselves to events where the
maximum amplitude occurs in the central BaF2 module.
While the incident neutron energies have to be obtained
by unfolding the kinetic energies Ecalc calculated with the
time-of-flight information of the central module, the incident photon energies can be directly derived from the measured total cluster amplitude by application of an additional
correction factor of 1.15. This correction accounts for the
energy leakage of the cluster and for the small difference
between the light output observed for cosmic-ray particles
and that of a genuine electromagnetic shower.
In Fig. 7 we compare the amplitude spectra for the central module and the whole BaF2 cluster. Selecting the same
four incident neutron energies already studied in Fig. 4 we
also illustrate the effect of the central hit condition. The increasing energy leakage from the central BaF2 module into
the surrounding detectors with increasing neutron energy is
evident. In accord with the growing energy deposit in the
A
1000
1500
E,,E,,, [MeVI
Fig. 8. The mean multiplicity of responding BaF2 modules in the
cluster of seven modules for photons and neutrons as a function
of the incident energy ,!$ and Ekin, respectively. For consistency,
a common threshold of La,, = 9 MeVs, was required for the individual amplitudes in both the cases. The events correspond to
central hits, as selected by the analysis condition that the central
module shows the maximum amplitude. GEANT3 results are given
by the solid lines.
outer modules the experimental multiplicity for the neutroninduced shower rises slowly with the primary neutron energy
and reaches a plateau of about 2.2 above a neutron energy
of 1OOOMeV. On the other hand, the multiplicity for the
electromagnetic shower increases rapidly with the energy
of the primary photon and reaches a value of 4.0 already
at a photon energy of 500MeV. The development of the
hadronic and electromagnetic showers with increasing incident energy is shown in Fig. 8.
The results of the simulations for the electromagnetic
shower, if analyzed according to the experimental conditions, are in good agreement with the measurements up to
the highest detected primary photon energy of 700 MeV. The
neutron-induced shower, however, is described only qualitatively by the GEANT3 simulations.
The probability to observe a given multiplicity of responding BaF2 modules as a function of the incident energy
is shown in Figs. 9 and 10 for neutrons and photons, respectively. The most probable multiplicity for the neutroninduced shower is one up to incident energies of 500 MeV,
and two above. Tbe situation is completely different for the
photon-induced electromagnetic shower. At 500 MeV incident energy the dominant multiplicity is already four. Moreover, the probability of events with multiplicity one drops
below 10% at photon energies of 200MeV and becomes
negligible for energies in excess of 500 MeV.
339
V. Wagner et al. I Nucl. Instr. and Meth. in Phys. Rex A 394 (1997) 332-340
Mul= 5
Mul = 6
3
0.6
t
0.4
0.2
0
0
1000
500
1500
E,,, [MeVI
Fig. 9. The probability to observe a given multiplicity (MUL)
of BaFz modules in the cluster detector of seven modules
(& = 9 MeV,) for neutrons as a function of the incident neutron
kinetic energy Ekin. GEANT3 results are given by the solid lines.
,x
..=
1
a
*
_\ Mul= 1
5!
,
Mul = 4
for incident neutrons with energies between 100 and
1300 MeV.
The response in pulse height is very poor. Neutrons
with sharp energy give rise to continuous spectra that
drop nearly exponentially and, to first order, the detection efficiency depends exponentially
on the electronic
threshold applied to the signal output. By extrapolation
the efficiency at zero threshold was determined. It reaches
a plateau of 30% at energies above 800MeV for the detector under study which has a length of 25 cm. Also, the
multiplicity response of the detector is weak. For incident neutron energies of several hundred MeV, the energy
leakage into neighboring modules becomes significant, but
remains substantially less than for photons with the same
energy.
The results of GEANT3 simulations
are in qualitative agreement with the measurements.
A quantitative
comparison, however, shows that the absolute efficiencies
are overpredicted by a factor of 1.3 -1.5 in the energy range
up to 750MeV, increasing to values of almost 3 for the
highest incident energies around 1300MeV. At a given
incident neutron kinetic energy the factor is higher for
higher electronic threshold. In addition, we find that the
multiplicity of the neutron-induced hadronic shower, which
is a measure of the lateral expansion of the shower, is
underestimated.
Taken together, these observations seem
to indicate that the GEANT3 simulations of the neutron
response of BaF2 scintillator based on the FLUKA package
underpredict the spatial development of the neutron-induced
shower.
Acknowledgements
This work was supported by the Grant Agency
Czech Republic under contract No. 202/93/l 144.
0.4
0.2
of the
References
[I] S. Kubota, T. Motobayashi, M. Ogiwara et al., Nucl. Instr.
0
0
100
200
300
400
500
600
700
800
E, [MeVI
Fig. 10. The probability to
of BaFz modules in the
(&,r = 9 MeVpe) for photons
energy E:,. GEANT3 results
observe a given multiplicity (MUL)
cluster detector of seven modules
as a function of the incident photon
are given by the solid lines.
and Meth. A 285 (1989) 436.
[2] T. Matulewicz, E. Grosse, H. Emling et al., Nucl. Instr. and
Meth. A 274 (1989) 501,
[3] R.A. Kryger, A. Azhari, E. Ramakrishnan et al.. Nucl. Instr.
and Meth. A 346 (1994) 544.
[4] A. Kugler, V. Wagner, M. Pachr et al., Phys. Lett. B 335
(1994) 319.
[5] R. Novotny, 1EEE Trans. Nucl. Sci. NS-38 (1991) 379.
[6] A.R. Gabler, M. Fuchs, B. Krusche et al., Nucl. Instr. and
Meth. A 346 (1994)
168.
5. Conclusions
[7] A. Gobbi et al., Nucl. Instr. and Meth. A 324 (1993) 156.
[8] R. Novotny, R. Riess, R. Hingmann et al., Nucl. Instr. and
Meth. A 262 (1987) 340.
We have measured both the response function and
the detection efficiency of a cluster of seven BaF2 modules
[9] M.E. RBbig, Workshop
on Physics
Schiermonnikoog,
1990, p. 176.
Related
to
TAPS,
340
V. Wagner et al. I Nucl. Instr. and Meth. in Phys. Res. A 394 (1997) 332-340
[lo] S. Ciejacks, M.T. Swinhoe, L. Buth et al., Nucl. Instr. and
Meth. 192 (1982) 407.
[ll] W.B. Amian, Nucl. Instr. and Meth. A 281 (1989) 353.
[ 121 J. Cugnon, Phys. Rev. C 22 (1980) 1885; J. Cugnon, Nucl.
Phys. A 387 (1982) 191~.
[13] R. Bnm, M. Hansroul, J.C. Lassalle, GEANT3 User’s Guide,
CERN DD/EE/84- 1, revision 1994.
[14] T. Matulewicz, Nucl. Instr. and Meth. A 325 (1993) 365;
G. Lanzanc, A. Pagano, S. Urso et al., Nucl. Instr. and Meth.
A 312 (1992) 515.