You are on page 1of 7

SCIENCE CHINA

Physics, Mechanics & Astronomy


• Research Paper • August 2010 Vol.53 No.8: 1453–1459
doi: 10.1007/s11433-010-4036-8

A comparison of n-γ discrimination by the rise-time and


zero-crossing methods
YAN Jie1,2*, LIU Rong2, LI Cheng1, JIANG Li2 & LU XinXin2
1
Department of Modern Physics, University of Science and Technology of China, Hefei 230026, China;
2
Institute of Nuclear Physics and Chemistry, Chinese Academy of Engineering Physics, Mianyang 621900, China

Received December 3, 2009; accepted January 5, 2010; published online June 30, 2010

The n-γ discrimination performance of two experimental arrangements based on the rise-time method and the zero-crossing
method was compared for a 50.8 mm-diametered and 50.8 mm-high BC501A liquid scintillator coupled to a 50.8 mm-diame-
tered 9807B photomultiplier in this work. The low energy limitation of the detected neutron with different detector high
voltages and the figure of merit of the n-γ discrimination in four neutron energy regions (1–2 MeV, 0.75–1 MeV, 0.5–0.75
MeV and below 0.5 MeV) were studied by using the Am-Be neutron source. Under a time statistical model of the
photoelectron emission process in scintillation counters, the intrinsic capability of the n-γ discrimination performance under the
optimal condition was evaluated. The experimental results of the zero-crossing method demonstrate a better n-γ. discrimination
performance than those of the rise-time method, which is consistent with the calculated results.

rise-time method, zero-crossing method, n-g discrimination, time statistics analysis, BC501A scintillation detector
PACS: 29.40.Mc, 07.05.Tp

1 Introduction discrimination results shown there only revealed an integral


effect above a certain neutron energy arising from the fact
Organic liquid scintillation detectors of type NE213 and that they just considered the low bias setting without a high
BC501A are widely used in the integral benchmark experi- energy limitation. In this case, the high energy neutron and
ments for the nuclear data validation by using the neutron gamma concealed the actual n-γ discrimination capability of
time-of-flight method for its high detection efficiency and the low energy. Besides, some statistical models have been
excellent n-γ pulse shape discrimination (PSD) properties built and studied since 1950s [8–13] not only for the
evaluation of scintillation materials but also for the optimi-
[1–4]. The most common n-γ PSD electronic circuit ar-
zation of PSD parameters and the comparison of PSD
rangement is based on the zero-crossing method. Some
methods. However, there is no report about the comparative
commercial pulse-shape analyzer (PSA) modules based on
the rise-time method can serve this function. For the per- and quantitative study of the n-γ discrimination performance
between the rise-time method and the zero-crossing method
formance of the n-γ discrimination impacts on the accuracy
focused on the low energy region.
of the neutron energy spectrum measurement, and for de-
For the neutron energy region above 2 MeV, as the decay
tectors with different sizes [5,6], comparisons of n-γ dis-
time of the pulse shape produced by neutron and gamma
crimination by zero-crossing and digital charge comparison
radiation has a large fraction of difference, it is relatively
methods [7] have been made in the past. However, the n-γ
easy to perform the n-γ discrimination. But in the low en-
ergy, especially below 1 MeV where the number of photo-
*Corresponding author (email: yanjie@mail.ustc.edu.cn) electron is small and statistical fluctuations are most often

© Science China Press and Springer-Verlag Berlin Heidelberg 2010 phys.scichina.com www.springerlink.com
1454 YAN Jie, et al. Sci China Phys Mech Astron August (2010) Vol. 53 No. 8

the predominant factor, it is difficult to separate the neutron while the other is used to obtain the recoil proton energy
and gamma with a high confidence level. spectrum for the selection of neutron energy regions.
In this paper, the n-γ. discrimination performance of two
n-γ discrimination experimental arrangements based on the
rise-time method and zero-crossing method was compared.
3 Experimental details
The energy of interest was below 2 MeV with four energy
regions 1–2 MeV, 0.75–1 MeV, 0.5–0.75 MeV and below To make the selection of the interested neutron energy re-
0.5 MeV from the Am-Be neutron source. Moreover, the gions, we first had to take the energy linear calibration of
statistical model described by Kuchnir [9] and Cao [13] was the multi-channel analyzer (MCA) for the recoil proton en-
adopted to evaluate the experimental results. ergy spectrum measurement by using the channel number of
the half height of the Compton edge of 137Cs monoenergetic
gamma ray source as the maximum energy recoil electron
2 Experiment setup with assumption that the energy linear of the MCA is good
enough. Because of the nonlinear response of the scintillator
The comparative study was carried out with the scintillation as a function of neutron energy, the selection of energy re-
detector composed of a 50.8 mm in diameter and 50.8 mm gions was imposed on the electron, and the light output
high BC501A liquid scintillator coupled to a 50.8 mm di- function for transforming the electron energy to neutron
ameter 9807B photomultiplier with silicon oil. The liquid energy was taken from Klein and Neumann [17], as shown
scintillators of type BC501A are identical to early NE213 in Figure 2. The neutron energy of interest (1–2 MeV,
scintillators in both design and chemical composition [14], 0.75–1 MeV, 0.5–0.75 MeV and below 0.5 MeV) was ob-
and the photomultiplier is designed with a good sin- tainable by calculating and adjusting the low-level dis-
gle-electron resolution (SER) and high pulsed linearity as a criminator and upper-level discriminator of the PSA (Figure
plug-in replacement for the RCA 8575 [15]. 1(a)) or of the TSCA (Figure 1(b)) associated with the value
The block diagrams of the two n-γ discrimination ex- of the neutron energy. To quantify the n-γ discrimination
perimental arrangements with the recoil proton energy in- capability in a given energy region, a figure of merit (FOM)
formation added are presented in Figure 1. Both of them [18]:
were established by a set of ORTEC electronic instruments.
One is based on the rise-time method by using the commer- Pn − Pγ
FOM = , (1)
cial PSA module ORTEC 552 [16], while the other is based FWHM n + FWHM γ
on the typical zero-crossing method. The dynode signal
from the scintillation detector for each of the two circuits is where Pn and Pγ are the channel numbers corresponding to
divided into two branches after being amplified by the the peaks of the n-γ discrimination spectrum for neutron and
Preamplifier. One branch is used for the n-γ discrimination gamma respectively, FWHMn and FMWHγ are their full

Figure 1 The block schemes of the two n-γ discrimination circuits. (a) is based on the rise-time method and (b) is based on the zero-crossing method.
YAN Jie, et al. Sci China Phys Mech Astron August (2010) Vol. 53 No. 8 1455

Figure 4, respectively. It is shown that in the energy region


1–2 MeV, both of the two methods exhibit excellent n-γ
discrimination. However, along with the decrease of the
neutron energy, there is more overlap of the n-γ discrimina-
tion spectra. For the rise-time method with neutron energy
below 0.5 MeV, it is hard to see the peaks of the neutron
and gamma because they are superposed on each other en-
tirely. In contrast, for the zero-crossing method, although
the n-γ discrimination is better than the former in each en-
ergy region and the peaks of the neutron and gamma are
distinct, the overlap is too serious to obtain good n-γ dis-
crimination in the case of neutron energy below 0.5 MeV,
Figure 2 Light output function L(E) for electrons, protons and alpha as summarized in Figure 4. For this reason, the value of the
particles as determined for an NE213 scintillation detector, 5.08 cm in FOM of the zero-crossing method in each energy region is
diameter and length (after Figure 2 of Klein and Neumann [17]). larger than that of the rise-time method, and the data fitting
with two Gaussion functions can not be performed for the
width at half maximum. All of them can be obtained by rise-time method in the energy region below 0.5 MeV. The
fitting the n-γ discrimination spectrum with two Gaussian comparison of the FOM for the two experimental arrange-
functions. ments in four energy regions is shown in Figure 5.
As an example, in the case of high voltage of −1900 V,
the Compton recoil electron spectra of 137Cs monoenergetic
gamma ray source used to perform the energy calibration 4 Time statistics analysis
and the n-γ discrimination spectra in the four neutron energy
regions for the two methods are shown in Figure 3 and Because both of the two n-γ discrimination methods utilizes

Figure 3 The Compton recoil electron spectra of 137Cs used to perform the energy calibration in the case of high voltage of −1900 V.

Figure 4 Comparison of the n-γ discrimination spectra in four neutron energy regions in the case of high voltage of −1900 V.
1456 YAN Jie, et al. Sci China Phys Mech Astron August (2010) Vol. 53 No. 8

definition of F(t) in eq. (3) is expressed by


t
F (t ) = ∫ p(t ′)dt ′. (4)
0

As the total photoelectron number n is proportional to the


energy deposition of radiation inside the scintillator, and the
number of photoelectrons per keV energy loss in the scin-
tillator for BC501A is about 1.7/keV [9], with the assump-
tion that the ratios of photons to neutrons in calculations are
to be one, the calculation results can predict the intrinsic n-γ
discrimination to be compared with the experimental results.
The corresponding parameters for the rise time distribution
calculation are listed in Table 1.

Figure 5 Comparison of the Figure of merit (FOM) in four neutron


energy regions in the case of high voltage of −1900 V. 5 Results and discussion

5.1 The relations of the lowest energy of the detected


the information of the pulse rise time, and the rise time is
neutron with different detector high voltages
proportional to the scintillation decay time, a statistical
model that describes the probability distribution of the rise In our experiment, the high voltage of the detector was
time introduced by Kuchnir and Cao. Under this model, the changed from −1400 V every other 100 V to −1900 V.
intrinsic performance of the n-γ discrimination for the two Since the neutron energy of interest is below 2 MeV, the
methods is evaluated and the experimental results are com- low energy limitation of the detected neutron is an impor-
pared. Due to their close implication to the problem we are tant criterion to compare the performance of the two n-γ
dealing with, the main conclusions are reported here; see discrimination methods. The relations of the lowest energy
refs. [9,13] for the detailed information. of the detected neutron with different detector high voltages
From an analytical point of view, the time profile of the are shown in Figure 6. The neutron energy is represented by
scintillation light pulse detected by the cathode of the pho- the equivalent electron energy. For the zero-crossing
tomultiplier in the case of NE213 and BC501A can be in- method, the lowest energy of the detected neutron is de-
troduced as the form: creased with the increase of the high voltage, and when the
high voltage exceeded as much as −1600 V, it is lower than
1 ⎛ q1 − t /τ1 q2 − t /τ 2 ⎞
p(t ) = ⎜ e + e ⎟, (2)
Q ⎝ τ1 τ2 ⎠
where Q is the total light output used to normalize its area to
1 (proportional to total energy deposition), and qi and τ i
(i=1, 2) represent the light yield and the decay time constant
of fast and slow component, respectively. The probability
distribution of the time at which a pulse reaches the fraction
x of its maximum, is given by

e − n n n [1 − F (t )]n − j [ F (t )] j −1 p(t )
Pj , n (t )dt = dt , (3)
( j − 1)!(n − j )!

where n is the total number of emitted photoelectrons pro-


duced by a monoenergetic radiation, which is assumed to be
Poisson distribution, j is the number of photoelectrons cor- Figure 6 The relations of detected neutron low energy limitation with
responding to the fraction x. It is apparent that j = nx. The different detector high voltages.

Table 1 Characteristic parameters of the BC501A scintillator used in our calculation (from ref. [13])

Scintillator Excitation τ1 (ns) τ2 (ns) q1 q2


n 3.94 47.1 0.527 0.473
BC501A
γ 3.36 24.6 0.556 0.444
YAN Jie, et al. Sci China Phys Mech Astron August (2010) Vol. 53 No. 8 1457

that of the rise-time method. For the rise-time method, the 5.3 The calculated results of the n-γ discrimination
lowest energy of the detected neutron is almost a constant from the time statistical analysis
close to 40 keV electron energy after the high voltage ex-
ceeded −1500 V. According to the time statistical analysis in sec. 4, computer
codes based on the Mathematica software system are writ-
ten to implement the calculation procedure. For a fraction x
5.2 The relations of the FOM with different detector of the maximum equal to 0.9 and for the total number of
high voltages in different neutron energy regions photoelectrons n in the range corresponding to the neutron
energy, the calculated results of the probability distribution
Figures 7(a)–(d) are the results of the relations of the FOM of the rise time for photon and neutron-induced scintillation
for the two n-γ discrimination methods with different de- pulses in our detector are shown in Figure 8. The Compari-
tector high voltages in different neutron energy regions. In son of the calculated results with the experimental results of
each energy region, the values of the FOM of the two n-γ the two methods is also shown in Figure 9, where the values
discrimination methods are almost the same along with the of the FOM of the experimental results in each energy re-
high voltage varying from −1400 to −1900 V, whilst the gion are the average value of different high voltages. In
value of the FOM of the zero-crossing method is larger than comparison, the rise-time method does not make a good n-γ
that of the rise-time method. In the energy region below 0.5 discrimination in each energy region, and the experimental
MeV, it is hard to carry out the fitting with Gaussian func- results of the zero-crossing method are close to the calcu-
tion, for the time distributions of neutron and gamma are lated results, whilst the value of the FOM of the calculated
superposed on each other for the rise-time method, and this results is slightly higher. The discrepancies evidently come
is why only the result for the zero-crossing method is dis- from four reasons. First, the parameters corresponding to
played in Figure 7(d). Furthermore, in this energy region, the decay of the scintillation light in our calculation are
because the lowest energy of the detected neutron is down from ref. [13] rather than the actual information of our de-
to about 0.1 MeV when the high voltage is −1900 V, a com- tector. Second, as the value of the transit time distribution of
plementary measurement in the neutron energy region of the PMT can not be taken from the specifications of the
0.25–0.5 MeV was performed and the value of FOM is PMT, the influence contributed to this factor is ignored in
close to 1 as shown in Figure 6(d) marked by square. our calculation. Third, the electronic noise distributions of

Figure 7 The relations of the FOM with different detector high voltages in different neutron energy regions. (a) 1–2 MeV; (b) 0.75–1 MeV; (c) 0.5–0.75
MeV; (d) below 0.5 MeV.
1458 YAN Jie, et al. Sci China Phys Mech Astron August (2010) Vol. 53 No. 8

clearly a superiority of n-γ discrimination method based on


the zero-crossing method compared to the rise-time method.
The effective separation between neutron and gamma is
down to about 0.25 MeV with the value of FOM close to 1
for the zero-crossing method.
Secondly, for the rise-time method, because the mini-
mum input threshold for the lower-level discriminator of the
PSA is 40 mV [16], the lowest energy of detected neutron
can only reach about 40 keV of equivalent electron energy
while the zero-crossing method can reach about 5 keV of
equivalent electron energy. An optimization work for the
design of electronic circuit of the PSA is needed in the fu-
ture for the specialized application in n-γ discrimination
Figure 8 Calculated rise-time distributions of the photon and neutron-in- below 1 MeV.
duced scintillation pulse in a BC501A in the four energy regions. Finally, without considering the influence of the transit
time spread of the PMT and the electronic noise of the elec-
tronic circuits, the time statistical analysis has provided a
precise evaluation of the intrinsic performance for these two
n-γ discrimination methods. The experimental results of the
zero-crossing method are closer to the calculated results
than that of the rise-time method. The reasons for these dis-
crepancies between calculated results and experimental re-
sults were discussed in sec. 5.3. In the neutron energy
measurement, if the low energy specifically below 1 MeV is
concerned, from the FOM point of view, the zero-crossing
method arrangement is a much better choice than the
rise-time method.

The author would like to thank Prof. Cheng Yuan for the fruitful discus-
sions and comments on the details of the experiment arrangement, and
Wang Mei for the use of 137Cs gamma ray source and Am-Be neutron
Figure 9 Comparison of the FOM of the calculated results and the ex- source in the experiment.
perimental results.

the two experimental arrangements are not considered. 1 Wong C, Anderson J D, Brown P, et al. Livermore pulsed sphere
program: Program summary through July 1972. UCRL-51144, Rev. I
Fourth, as the authors pointed out in ref. [9], it could be an 2 Oyama Y, Maekawa H. Spectral measurement of angular neutron
error to assume that the relation between the average actual flux on the restricted surface of slab assembles by the Time-of-Flight
number in an individual pulse is given by Poisson statistics. method. Nucl Instrm Method A, 1986, 245: 173–181
Consequently, our calculated results provide the intrinsic 3 Tomoo N, Kazusuke S. Integral experiments for fusion reactors. Nu-
n-γ discrimination performance under the ideal condition. clear Data for Science and Technology, 1988 MITO: 171–178
4 Ichihara C, Kimura I, Yamamoto J, et al. Measurement and analysis
of leakage neutron spectrum from a spherical pile of silicon with in-
cident 14 MeV neutrons. J Nucl Sci Technol, 2007, 44(1): 29–35
6 Conclusions
5 Moszyński M, Costa G J, Guillaume G, et al. Study of n-γ discrimi-
nation with NE213 and BC501A liquid scintillators of different size.
The low energy limitation of the detected neutron as a func- Nucl Instrm Method A, 1994, 350: 226–234
6 Naqvi A, Nagadi M M, Rehman K, et al. Performance comparison of
tion of the detector high voltages and the FOM of the n-γ
NE213 detectors for their application in moisture measurement. Appl
discrimination by rise-time and zero-crossing methods in Radiat Isot, 2000, 53: 745–754
four neutron energy regions below 2 MeV are compared in 7 Wolski D, Moszyński M, Ludziejewski T, et al. Comparison of n-γ
this paper. In conclusion, we should like to emphasize the discrimination by zero-crossing and digital charge comparison meth-
following general points with regard to the n-γ discrimina- ods. Nucl Instrm Method A, 1995, 360: 584–592
tion performance for the rise-time and zero-crossing meth- 8 Post R E, Schiff L I. Statistical limitations on the resolving time of a
scintillation counter. Phys Rev, 1950, 80: 1113–1120
ods.
9 Kuchnir F T, Lynch F J. Time dependence of scintillations and the
Firstly, although these two methods make use of the rise effect on pulse-shape discrimination. IEEE Trans Nucl Sci, 1968,
time information of the pulse, the presented study shows NS-15(9): 107–113
YAN Jie, et al. Sci China Phys Mech Astron August (2010) Vol. 53 No. 8 1459

10 Frank J, Lynch. Basic limitation of scintillation counters in time uid scintillation spectrometers. Radiat Prot Dosim, 2003, 107:
measurements. IEEE Trans Nucl Sci, 1975, NS-22: 58–64 95–109
11 Gioacchino R. Time statistics of the photoelectron emission process 15 9807B data sheet. http://www.saetong.com/products/ETL/forms/pdf/
in scintillation counters. Nucl Instrm Method A, 1993, 335: 121–128 9807B.pdf
12 Gioacchino R. An analytical approach to the evaluation of the pulse 16 Model 552 Pulse-Shape Analyzer and Timing Single-Channel Ana-
shape discrimination properties of scintillators. Nucl Instrm Method lyzer Operating and Service Manual. ORTEC@ Part No.733460, 2002
A, 1995, 354: 389–399 17 Klein H, Neumann S. Neutron and photon spectrometry with liquid
13 Cao Z, Miller L F. Evaluation of pulse shape discrimination per- scintillation detectors in mixed fields. Nucl Instrm Method A, 2002,
formance of scintillation materials and PSD methods by using statis- 476: 132–142
tical models. Nucl Instrm Method A, 1998, 416: 32–44 18 Knoll G F. Radiation Detection and Measurements. New York: John
14 Klein H. Neutron spectrometry in mixed fields: NE213/BC501A liq- Wiley & Sons Press, 1976

You might also like