2.1 Introduction and result of experiment

Using a CsI detector is the easiest way to test a detector using standard radioactive sources. However, the energy of γ-rays produced by the radioactive sources are usually very low. Although the efficiency and energy resolution of the detector for some energy can be obtained directly, the range of energy calibration is relatively small. Therefore, simulation is required for extrapolating the detector performance to the high energy region.

Three radioactive sources, ¹³⁷Cs, ⁶⁰Co, and ²²⁸Th were selected for the experiment. γ-ray of 2.614 MeV energy can be obtained from ²²⁸Th so that energy range of detector calibration is between 662 keV and 2.614 MeV. Since our main goal is the detection of high-energy γ-ray, the efficiency of γ for energy less than several hundred keV is not studied. Thus, the ¹⁵²Eu source is not used as an experimental radioactive source.

The CsI(Tl) detector used in this study was developed by the Shanghai Institute of Applied Physics, Chinese Academy of Sciences. It is a sub-project of the Thorium-based Molten Salt Reactor (TMSR) pilot project. The initial aim of this detector is to measure the neutron capture reaction. A large number of CsI(Tl) detectors are installed around the experimental sample. For several MeV γ-rays from neutron capture reaction, the main requirement for the γ detectors is high detection efficiency. Since the CsI crystal is very long, it also provides favorable conditions for measuring the high energy γ-ray.

The whole CsI(Tl) crystal detector is a hexagonal prism, which is 200 mm long, and the distance between parallel edges of intersection surface is 55 mm. The crystal is produced by the Shanghai Institute of Ceramics, Chinese Academy of Sciences, and is covered with 1 mm polytetrafluoroethylene and 2 mm thick aluminum as shielding materials. The photomultiplier tube connecting to one side of the crystal is R1828-01 (Hamamatsu company). [17]. Figure 1 shows the assembly diagram of the detector and a photograph of the detector.

Figure 1:Full-screen View

(Color online) Structure diagram of the CsI detector unit.

The main amplifier MSCF-16 is used in our test, and the signal is amplified and inputted to the ADC (CAEN v785) for A-D conversion and read out by the VME data acquisition system.

Fig.2 presents the γ spectra measured by the CsI detector. As we can see, the energy spectra of ¹³⁷Cs and ⁶⁰Co is clear, just one or two peaks, but the spectrum of ²²⁸Th has a number of peaks. The peak at 238.6 keV is high while the peak at 2.614 MeV is relatively very low, because the fractional ratio is relatively smaller and the detection efficiency of γ-ray decreases with the increase of energy. Despite this, the peaks are still identified clearly. Six peaks from the three sources are used to calibrate the γ energy signal of CsI. The energy calibration of the detector is presented in Fig.3, we can see good linear relationship between the energy and the channel number when the energy of γ-ray is below 2.6 MeV.

Figure 2:Full-screen View

(Color online) Measured γ-ray spectra of ¹³⁷Cs, ⁶⁰Co, and ²²⁸Th.

Figure 3:Full-screen View

(Color online) Linear fit between the channel number and the energy deposited in a CsI:Tl scintillation detector in an energy range from 200 keV to 3 MeV with a hamamatsu R1828-01 phototube and the HV setting to 1600 V.

2.2 Energy resolution

Energy resolution is an important capability of the detector. It is defined as

where E and FWHM represent the peak value and the full width of the γ energy spectrum [18]. The relative energy resolution of the detector is mainly related to the characteristics of the crystal itself and performance of the photomultiplier tube. The resolution of the CsI detector is correlated to the γ energy (E) [19,20]. The results obtained by fitting the experimental data reduce from 17% to 3.5% with gamma sources energy growth, as shown in Fig.4, The fitted results can be expressed by the following function

Figure 4:Full-screen View

Relative energy resolution for γ-rays in a CsI:Tl detector.

The uncertainties of the data are very small and negligible. This is consistent with some of the previous experimental results [21-23].

3.1 Detector construction model

The γ-ray emitted from radioactive sources used in the experimental measurements are isotropic in space. The crystal is a hexagonal prism type and its surrounding space is vacuum. Every radioactive source with a very small surface was located 5 mm in front of the CsI crystal. The γ-ray comes out from the source in an isotropic emission way, so the γ photons fly into the crystal from different angles. Because the efficiency of the detector is our first concern, other factors such as visible photon emitting from the energy deposit of the γ-ray, transmission process of visible photons in the crystal, and the surface features of CsI crystal, were not considered when we established the simulation environment.

3.2 Simulation results

GEANT4 toolkit plays the main role in simulating the whole energy deposition of γ-rays from the ¹³⁷Cs (662 keV) source and ⁶⁰Co (1173 keV, 1332 keV) source. Two γ-rays energies from the ⁶⁰Co source were used in the simulation with the real emission probabilities at the same time. Since the simulation is about the energy deposition of the γ-ray, the simulation process does not include the production, transmission, and collection of scintillation light in the crystal. The statistical error and fluctuation from electronic noise of the photon-read device was not considered neither [20]. So we have used the energy resolution obtained by fitting the experimental data in the simulation to show the dispersion of energy deposition. The results are the same as Fig.2. That means the simulation reproduces the test results of CsI detectors well. The simulated energy spectrum of ⁶⁰Co and ¹³⁷Cs are shown in Fig.5.

Figure 5:Full-screen View

Simulation results of γ in the CsI from ¹³⁷Cs (a), ⁶⁰Co (b).

By changing the energy of the γ-ray in GEANT4, we can simulate the effect of different energy, especially high energy, γ-ray hitting in the large volume CsI scintillation. A series of different energy spectra were obtained. As shown in Fig.6, the inset plots are the energy spectra without considering the energy resolution, while the full size plots are the energy spectra with the inclusion of the energy resolution effect [24].

Figure 6:Full-screen View

Simulation of four kinds of isotropic gamma sources energy spectrum without (inner) and with the broadening effect (outer).

It should be noted in Fig.6 that as the energy of the γ-ray increases, a single electron escape peak and double electron escape peak begin to appear. When the energy is only 661 keV, the energy spectrum has only one sharp full absorption peak. When the energy reaches 8 MeV, we can find that the simulation results have not only the full absorption peak, but also the single electron escape peak and the double electron escape peak with 0.511 MeV shift in energy. As the energy increases, proportion of the full absorption peak over a single electron escape peak decreases gradually. If the energy resolution is introduced, the results should be consistent with the results of the detector test. As shown in Fig.6, it can be found that as the energy increases, the spectrum broadening increases. When γ energy reaches 16 MeV, the full absorption peak, single electron escape peak, and double electron escape peak have been overlapped together.

The detection efficiencies of the single and double escape peak were extracted from the simulation results. We can see from Fig.7, the measured points were a little bit lower than the simulation results. The detection efficiency at low-energy is in quite good agreement with the simulation results. Meanwhile, the detection efficiencies at the high energy region are also predicted. The efficiency of a single escape peak is higher than the total absorption peak for the γ energies in the 5-15 MeV range, indicating that it is feasible to select it as the characteristic peak for efficiency study.

Figure 7:Full-screen View

The energy dependent photopeak absorption efficiencies for isotropic incidence simulated by GEANT4.The experimental data are shown by the stars.

The emission style of γ source in our simulation is isotropic. But in some cases, for example, a photon nucleus reaction, the γ-ray is a parallel or quasi-parallel beam. So we have simulated and drawn the efficiency line using the same parameter except that the incident direction of γ is perpendicular to the cross section of the crystal.

When the γ ray hits vertically on the CsI crystal, the effective range of γ in the crystal increases, and the probability of deposited energy inside the crystal increases, thus the efficiency will also increase. As Fig.8 shows, the CsI detector has a high detection efficiency up to 66% for 1 MeV γ and 3.8% for 20 MeV γ.

Figure 8:Full-screen View

The energy dependent photopeak absorption efficiencies for normal incidence simulated by GEANT4.

In addition, from the simulation results, single electron escape and double electron escape are similar to those of the total energy peak at 20 MeV. If the single electron escape peak and double electron escape peak can be taken into account [25], a higher efficiency can be obtained. Moreover, the simulation results show that when the energy of the γ-ray increased, the three peaks will be overlapped together. Further study is needed to improve the efficiency of the detector for high energy γ-rays.