2.1 Experiment

To investigate the performance of the selected lithium-glass scintillator, an experimental study was carried out. The cylindrical lithium-glass sample has dimensions of ϕ 40 mm × 4 mm. It has a total lithium content of 7.7 wt.% with ⁶Li enrichment of 93%. In the experiment, the glass was wrapped in aluminum foil to improve the light collection efficiency. The unwrapped end was mounted on the front window of a PMT (Philips XP2020) by silicone grease. A bias voltage of -1800 V was applied to the PMT. An oscilloscope (Agilent DSO-X3024A) with an interface developed using LabVIEW was used to perform data acquisition.

The single-photoelectron response of the XP2020 PMT was calibrated in advance. A light-emitting diode with a light wavelength of 395–400 nm, which was chosen according to the maximum scintillation light wavelength of 395 nm, was driven by a pulse generator as the calibration light source. We obtained a single-photoelectron response of 6.2× 10² fC/photoelectron.

The ²⁵²Cf neutron source at the IHEP, which has a neutron yield of approximately 1× 10⁶ n/s, was used to study the neutron response of the lithium glass. A diagram of the experimental layout is shown in Fig. 2. The source itself is surrounded by lead sphere shielding to suppress gamma rays. The neutron moderator layer consists of 35-cm-thick paraffin. A layer of borax with a thickness of 11 cm is placed as the outermost layer for dumping of the moderated neutrons. A neutron beam collimator tunnel with a diameter of 10 cm is prepared starting from the lead sphere and continuing through the neutron moderating and dumping materials. The lithium glass coupled with the PMT is positioned next to the collimator tunnel. Because most neutrons coming through the tunnel are fast neutrons, a cylindrical moderator with a thickness of 10 cm is placed in the tunnel to thermalize the neutrons in this experiment. To study the n-gamma discrimination capability of this glass, a ⁶⁰Co gamma ray source experiment was also carried out using the same experimental setup.

Figure 2:Full-screen View

Diagram of the ²⁵²Cf neutron source experiment for studying the performance of the lithium-glass scintillator.

2.2 Monte Carlo simulation

As a traditional choice of neutron beam monitor, the lithium-glass detector has some advantages over the gas detector. i) It can be processed into a needed shape easily. ii) The volume is smaller for the same detection efficiency. iii) No gas-maintaining accessories are required. At the same time, a well-known shortcoming is its excessive sensitivity to gamma rays. Along the neutron beam line of the SANS, the intensity of the associated gamma rays is high. Some of them are direct side products of the neutrons on the spallation target; the others come from neutron reactions with the materials around the spectrometer. To significantly reduce the neutron misidentification due to gamma rays and to establish a proper design, a simulation was performed using the Geant4 Monte Carlo simulation toolkit (version 10.1). The lithium glass that we use is similar to the commercial product GS20; hence, it was modeled with a composition of 18% Li₂O (⁶Li enriched to 93%), 4% MgO, 18% Al₂O₃, 57% SiO₂, and 3% Ce₂O₃ [11]. All the physical processes related to detection were added within the framework provided by Geant4. The high-precision neutron nuclear reactions model driven by cross section data was implemented to handle the neutron transport. The PENELOPE code embedded in Geant4 was used to simulate the electromagnetic progresses. The scintillation and light collection were modeled using the optical photon package of Geant4 [12]. The attenuation length of the scintillation light is ∼50 cm [13], which is much larger than the glass thickness. Consequently, the attenuation effect is not significant. The reflectivity of the aluminum foil wrapping was set to 70%. The other scintillating and optical parameters come from reference [14].

3.1 Charge spectra

Typical neutron and gamma-ray signals are displayed in Fig. 3(a). The corresponding charge spectra of the lithium glass for ⁶⁰ Co and ²⁵²Cf are shown in Fig. 3(b). In the spectrum of ²⁵²Cf, the left "peak" is attributed to the background on the basis of the experimental result obtained without a source. The right peak is the neutron peak. The light yield of a neutron, Y, can be obtained by the following equations:

Figure 3:Full-screen View

(Color online) (a) Typical signals of a neutron and a gamma photon. (b) Charge spectra of ²⁵²Cf neutrons and ⁶⁰Co gammas.

where Q_n is the average charge corresponding to a neutron, and Q_s is the charge response of a single photoelectron. The charge response of a neutron, Q_n, is 5.3×10² pC, which is derived from the peak position of the charge spectrum. The scintillation light collection efficiency, ε_c 80%, is acquired by a light transition simulation with the Geant4 toolkit, which has been verified in our previous work [14]. ${\overline{\epsilon }}_{\text{q}}$ is the average quantum efficiency of the PMT for the scintillation light; it is expressed as

where I(λ) is the normalized scintillation light spectrum, and ε_q(λ) is the quantum efficiency of the PMT for a wavelength λ. We did not measure the scintillation spectrum of this glass. The average quantum efficiency ${\overline{\epsilon }}_{\text{q}}$ is approximated by the quantum efficiency ε_q 20% at the wavelength of maximum emission (395 nm). Thus, the light yield of this scintillator, Y, for a neutron is determined to be 5.3×10³ photons/neutron, which is close to that of the commercial lithium glass GS20, 6×10³ photons/neutron [15].

A thermal neutron is absorbed in lithium glass by the following reaction:

The charged ions (the triton and alpha) are emitted with initial kinetic energies of 2.73 and 2.05 MeV, respectively. Their ranges in lithium glass are 40 and 7 μm, respectively, which are far smaller than the thickness of the glass sample. Therefore, the energy deposited by the neutron-produced ions does not generate additional smearing of the neutron spectral peak. The peak width indicates an energy resolution of 16%. In the spectrum of ⁶⁰Co, the left "peak" comes from the background cut by the threshold. The Compton edges of the two energy lines at 1.173 and 1.332 MeV merge and form a peak-like structure (the right "peak" in the spectrum at ∼300 pC). Lithium glass is composed of light elements, and the tested sample is thin; therefore, the peak-to-Compton ratio is low. As a result, the photoelectric peaks are merged by the Compton plateau due to the energy resolution. The ⁶⁰Co spectrum from the Monte Carlo simulation agrees well with that shown in Fig. 3(b), where the Compton-edge "peak" position is ∼0.9 MeV. The neutron peak at ∼530 pC is distinguished from the gamma spectrum. However, many events overlap each other owing to the gamma sensitivity of the lithium glass. The neutron peak corresponds to an electron-equivalent energy of 1.6 MeV [16]. The positions of the neutron peak and the Compton-edge "peak" together verify the linearity of the PMT gain.

3.2 Gamma sensitivity

Decreasing the thickness of the glass plate is an effective way to reduce the gamma sensitivity of the neutron monitor. The electron-equivalent energy of a thermal neutron E_ne 1.6 MeVee [16]. An electron with an energy of 1.6 MeV has a range of ∼2.9 mm in lithium glass. It has a high probability of leaving a glass plate with a thickness smaller than its range before exhausting its kinetic energy. Therefore, the energy deposited by a gamma ray is limited by the thickness of the lithium glass regardless of the energy of a gamma photon. The gamma sensitivity ξ is introduced to evaluate the n-gamma discrimination ability. It represents the probability that a gamma signal with a specific energy is mistaken for a neutron signal. Its value can be derived using the formula [17]

where C_γ is the number of gamma rays mistaken for neutron signals, and C_t is the total number of detected gamma rays. According to the experimental result, the energy resolution of the thermal neutron peak given by this type of lithium glass is 16%. A threshold for eliminating gamma events with E_ne- FWHM=1.344 MeV is implemented. Thus, gamma rays depositing energy above the threshold are expected to be mistaken for neutrons.

To validate the code, thermal neutrons and gamma rays with an energy of 1.332 MeV from ⁶⁰Co are generated to strike a glass scintillator sample 4 mm in thickness in the simulation. The energy spectra are shown in Fig. 4(a). They are consistent with the experimental result. When the glass thickness decreases to 2.0 mm, there is no apparent change in the neutron spectrum. However, the neutron-gamma separation becomes larger. Under this condition, even higher energies (e.g., 5 and 10 MeV) lead to a further decrease of the overlap between the neutron spectrum and gamma spectrum. The systematic results for the gamma sensitivity are given in Fig. 4(b). When the thickness is 0.5 mm for a gamma energy of 1.5 MeV, the sensitivity of the lithium glass reaches a magnitude of 10^-6. This value is close to that of a well-behaved ³He detector with an appropriate threshold [17].

Figure 4:Full-screen View

(Color online) (a) Electron-equivalent energy (E_e) spectra of neutrons and gamma rays given by the simulation. (b) Gamma sensitivity of gamma rays with different energy (E_g) in multiple-thickness scintillators.

3.3 Lithium glass selection

The width of the pulse output from the lithium glass is 200–300 ns, as shown in Fig. 3(a). Accordingly, the counting capability of the monitor based on the lithium glass is approximately 3.3×10⁶s^-1 if we ignore the dead time of the electronics. Therefore, the detection efficiency of the monitor should be kept below 5.7% considering the beam intensity. The simulated neutron absorption efficiency of scintillators with different thicknesses is shown in Fig. 5. The neutron spectrum at the SANS is supposed to peak at the thermal energy (1.8 angstrom). Thus, we consider a detector design based mainly on thermal neutrons. The thermal neutron absorption efficiency is approximately 59% for 0.5-mm-thick lithium glass. To reduce the scintillator thickness is an effective and direct method to decrease the efficiency. The minimum thickness that the vendor of this type of lithium glass can supply is 0.1 mm. At the same time, thinner glass also has a lower gamma sensitivity. However, the monitor is supposed to work under a vacuum environment and may be moved frequently into and out of the beam line. When the thickness is less than 0.5 mm, the glass tends to be deformed or even break under the strain caused by the difference in pressure. If defects are created during operation, the measurement accuracy will be affected. Thus, the use of low-enriched or natural (⁶Li) lithium glass is an appropriate option that yields acceptable gamma sensitivity without mechanical problems. The natural enrichment of ⁶Li is 7.4%, and that of the glass used in the experiment is 93%. Accordingly, a 0.5-mm-thick lithium glass with natural lithium has a neutron detection efficiency of approximately 4.7%, which satisfies both the gamma sensitivity and efficiency requirements.

Figure 5:Full-screen View

(Color online) Neutron detection efficiency versus neutron wavelength for different scintillator thicknesses.