Nondestructive testing is a key feature of neutron imaging [1
]. The neutron contrast modality depends on the specific nuclear core structure, enabling the analysis of both heavy and light elements, as well as differentiation between isotopes. Hence, neutron imaging has been established as an efficient technique in materials science, archaeology and engineering, whenever X-rays do not provide enough contrast. In recent times, one focus of neutron imaging is time-resolved imaging in the sub-second regime [2
]. Exposure times in this regime pose challenges with respect to the available neutron flux, the camera system, as well as the decay time of the scintillation materials, which are an essential component of almost all detection systems for neutron imaging. A shorter exposure time causes, at a particular neutron flux, less photons to be detected in one image, leading to increased Poisson noise in the images. Furthermore, the ratio between the generated signal and the thermal noise in the camera has to be high enough to be able to distinguish signal from noise. The decay time of the light emission of scintillation screens influences the following images, as the light emission generated in one frame may still be emitted during the exposure of the subsequent frame i.e., generating ghost images. This degrades the quality of the measured data. Rapidly decaying scintillation screens are needed to suppress this effect. In our experiments, one of the standard scintillation screen mixtures (ZnS:Cu/6
LiF) shows a relatively slow decay time of approximately
to 1% of the light yield after the end of irradiation with neutrons. It should be noted that this result differs from previously recorded results, which suggest a decay time of 85 μs to 10% [5
] in a ZnS/6
LiF scintillation screen. We expect this deviation to be a result of the different excitation radiations used in the experiments. In the discussion, we give a more detailed explanation of the origin of this discrepancy.
While various studies of the decay of neutron scintillation screens have been performed in the past [6
], those studies investigated the decay in the μs-range. In recent times, the focus of scintillation screen development has been on high spatial resolution, where multiple studies have been performed [8
]. New scintillation screens specifically for
- discrimination have also been studied [11
]. In these studies the decay time of the scintillation screens has been examined as well. In this publication, we characterize the time and flux dependent light yield characteristics of five new commercially available scintillation screen mixtures on a time scale of seconds and compare them with standard ZnS:Cu/6
LiF. The evaluated mixtures can be split into two categories. Four scintillation screens are 6
LiF-based while two are Gd-based. A detailed description of the material properties is given in Section 2.1
. The scintillation screen mixtures listed in Table 1
were provided by RC Tritec AG
] and were chosen because their scintillating material is expected to show a fast decay. To characterize each scintillation screen, we have measured its light yield during the first few seconds of irradiation with neutrons, as well as after the end of irradiation.
For ZnS:Cu/6LiF, we additionally analyzed the change of light yield for exposure to different neutron flux levels as well as for different scintillation screen temperatures. It must be noted that using only the light yield does not give a full characterization of the used scintillation screens. The conversion efficiency as well as the neutron detection efficiency were not studied in detail. Although we did not performed a study on the inherent limitations in spatial resolution of the scintillation screens, we do expect a behavior similar to standard scintillation screens i.e., the spatial resolution is primarily influenced by the thickness of the scintillation screen.
In this publication, we give a phenomenological description and analysis of the light yield change of the scintillation screens due to sudden flux changes. Here, the interesting process is not the absorption of the neutron by the scintillation screen, but the excitation of the scintillating material by the absorption products. In the 6LiF-based scintillation screens the excitation radiation is the same, hence we focus on those screens when comparing the light yield changes.
4. Discussion and Conclusions
In this paper, we presented our findings on five novel scintillation screen materials (ZnCd, YAG, ZnO, Li-Gadox, Gadox) shown in Table 1
as well as compared them to ZnS, which is one of the most common scintillation materials used in neutron imaging. The light yield of the 6
LiF-based scintillator screens tends to be generally higher than the light yield of the Gd-based scintillators, as can be seen in Figure 1
. ZnS shows the highest light yield and has been used as a reference for the other scintillators. Li-Gadox and Gadox show a high neutron absorption ((68.3 ± 3.2)% and (67.3 ± 2.4)%) but a low light yield ((5.8 ± 0.2)% and (3.8 ± 0.2)%) compared to the light yield of ZnS. ZnCd shows the second highest normalized light yield with (63.3 ± 2.5)%, while YAG has a normalized light yield of only (9.1 ± 0.2)%. The neutron absorption of ZnCd is (29.6 ± 1.9)%, whereas YAG absorbs (12.7 ± 2.9)% of the incoming neutrons. The higher absorption in the thinner Gd-based screens is expected, as Gd has a much higher neutron absorption cross section (49700
) than 6
Li ( 940
]. The low relative light yield of the Gd-based screens can be explained by the different daughter products of the conversion reactions of 6
Li and Gd. Gd produces a conversion electron with an energy of 29 keV to 191keV when capturing a neutron, whereas 6
Li produces an
-particle with 2055
and a 3
H-particle with 2727
. The lower amount of scintillation material, due to the smaller thickness of the Gd-based screens, as well as the lighter and less energetic particles generated by the neutron absorption leads to less light being emitted by Gd-based screens compared to 6
LiF-based screens. In contrast, the heavier and more energetic particles produced by 6
Li as well as the higher scintillation screen thickness leads to more energy being deposited in the scintillation material, typically causing higher light yields. Contrary to this trend, ZnO barely shows any measurable light yield and is therefore not suitable for neutron imaging.
Even though the basic scintillation mechanism of electron excitation is the same for all four scintillation materials in the 6
LiF-based screens, the specific mechanisms governing the probability of light emission are influenced by the specific crystal structure and are beyond the scope of this paper. A more detailed view upon the underlying scintillation processes is given by Getkin et al. [20
It should be noted that the higher neutron absorption of ZnCd does not necessarily indicate a higher detection probability of neutrons, as the Cd in this screen results in additional absorption. Although the absorption of neutrons in Cd produces
-rays, we expect no significant contribution to the light yield caused by
-rays. Typically, ZnS-based scintillation screens show low sensitivity to
-rays. As Spowart [21
] suggests that 103
photons are required to produce the same light yield as a single neutron.
Considering the buildup of light yield in the beginning of irradiation (first 1000
) shows that Li-Gadox, Gadox and ZnCd do not have an appreciable buildup of light yield, whereas ZnS displays an increase of light yield to 104%. YAG also shows no buildup, but suffers from decreasing light yield. The buildup curve of ZnS is thought to be a combination of three different processes. Typically, ZnS:Cu has intrinsic electron traps. In the beginning of irradiation a part of the electrons excited by the neutron capture process show no luminescence, but are captured by these traps. The electrons in the traps may decay later, showing delayed scintillation. Continuing neutron radiation causes an equilibrium between the charging and decaying of the traps, as well as the direct fluorescence [22
]. This initial equilibrium is reached roughly 60
after the beginning of irradiation, see Figure 2
a. Afterwards, ZnS shows a continuing decrease of light yield due to its relatively poor radiation hardness. Furthermore, the damage to the crystal structure caused by ionizing radiation also leads to the creation of color centers i.e., defects which act as electron or hole traps further reducing the light yield [20
]. The reduction in light yield continues during the whole experiment, as seen in Figure 3
. Over the 12
break in the long term measurements indicated by the dashed line B in Figure 3
, the color centers and defects in the crystal structure are partially annealed, which leads to a slightly increased light yield [26
]. An increase in temperature causes thermal quenching for ZnS:Cu, reducing the light yield, while the trend of decreasing light yield due to radiation damage continues [27
The neutron fluence dependent buildup of ZnS in Figure 2
b using different neutron fluxes and scintillation screen temperatures shows a slightly larger maximum relative light yield, i.e., a stronger buildup, for the high neutron flux compared to the low flux. Simultaneously, the relative light yield decreases faster with rising neutron fluence in the case of higher neutron flux. This indicates that the filling of traps and the creation of defects in the crystal structure is primarily neutron fluence dependent while the decay of traps is primarily time dependent. Hence, a higher neutron flux causes more traps to be filled simultaneously, increasing the peak buildup. At the same time, increasing the neutron flux leads to increased defect creation, indicated by the faster decline of relative light yield during irradiation with high flux [20
The increased relative buildup of the heated scintillation screen is presumably caused by the fact, that the number of intrinsic traps stays the same over an increase in temperature, while the direct fluorescence is quenched. However, we expect the life time of the electrons in the traps to be reduced as their decay may be thermally induced. Hence, the general shapes of the curves are similar, but the absolute buildup values originate from different starting conditions and are therefore not comparable. Heating of the scintillation screen seems to lead to a slightly slower decrease of light yield (≈ s−1) compared to the scintillation screen at room temperature (≈ s−1).
The light yield curve of ZnCd during constant neutron irradiation shows only a small decrease of light yield, indicating an increased radiation hardness. Here, the Cd in ZnCd is suggested to be a stabilizing agent reducing radiation damage by quenching the luminescence caused by the luminescence centers [20
]. The measurement break of 12
also leads to a small increase of light yield again indicating an annealing of crystal defects. The effect of quenching at higher scintillation screen temperatures is slightly stronger for ZnCd compared to ZnS.
The strong decrease of light yield in YAG is surprising, as YAG has been reported to be stable with respect to radiation [28
]. It needs to be noted that the radiation hardness test in [28
] was performed using a 3
proton beam, hence the results may not be directly comparable. YAG also shows no thermal quenching, which may be a result of a different crystal structure compared to the other two scintillation screens. Similar to the other screens YAG shows an increased light yield after the 12 h resting period.
After the end of irradiation ZnS shows an order of magnitude higher afterglow than ZnCd, while the afterglow of YAG, Li-Gadox and Gadox lies in the range between these two extremes. We note that the decay of the scintillation screens does not follow a single exponential decay, but is a superposition of various exponential decays. While the data curves may be fitted with multiple exponential functions, the resulting parameters would give no further information. Hence, no fitting of the data curves was performed. One should also note that due to the low absolute light yield of YAG, Li-Gadox and Gadox, the noise in the data becomes more pronounced. This noise is caused by -radiation hitting the detector as well as thermal noise in the camera. Incidentally, this noise may also be the reason for the flattening of the curves, as the constant noise may hide a slowly decaying light output. The leveling off of the curves occurs at different relative levels as the light yield before shut off of the neutron beam was different for every scintillation screen, while the noise level in the detector system was constant for every scintillation screen. This may also explain the crossing of the decay curve of ZnS with the curve of Li-Gadox at . The high afterglow of ZnS probably originates from radiation induced luminescence centers. As “de-trapping” is thermally induced, the increase of temperature leads to a faster decay. Furthermore the thermally induced annealing process reduces the amount of induced luminescence centers and hence reduces the afterglow.
As noted before, previous studies mostly covered the response of scintillation screens in the μs-range [6
]. Furthermore, these studies were performed at low flux sources (≈106
). Hence, the long decay time we evaluated may be hidden in the measurement noise of these studies, similar to our experiences with YAG, Li-Gadox and Gadox.
One important property of neutron scintillation screens, their spatial resolution, has not been touched upon in this study. However, we expect the scintillation screens to follow roughly the same trend of spatial resolution versus thickness as observed in standard scintillation screens offered by RC Tritec AG
]. Comparing the five new scintillation screen mixtures, the ZnCd mixture has the highest light yield, while the Gd-based screens have, with the exception of the 6
LiF-based ZnO screen, the lowest light yield. While the decay time of Li-Gadox and Gadox is slightly faster than that of ZnS, their low light yield per detected neutron increases the Poisson noise in the camera system [1
]. While this property can be overcome with increased exposure times, this is not feasible for high frame rate neutron imaging. While Gd-based scintillation screens are not suitable for high frame rate imaging, their high detection probability at low thicknesses makes them a good choice for high spatial resolution imaging. The non-existent buildup and low afterglow of YAG would make it well suited for high frame rate neutron imaging. In contrast, the continuously decreasing light yield as well as the generally low light yield compared to other 6
LiF-based scintillation screens make this mixture unattractive for neutron imaging. ZnCd has 63% of the light yield of the standard ZnS mixture. The buildup is in the range of the neutron flux fluctuation of the reactor. Furthermore, for ZnCd we measured the fastest decay of afterglow of all evaluated scintillation screens. ZnS decays to 1.98% of its previous light yield in 1
, while ZnCd decays to 0.14% in the same time. An advantage of ZnCd compared to other fast decaying scintillation screens [29
] is the high light yield per detected neutron, which is much higher than observed in Gd-based scintillation screens or in Li-glass scintillation screens [1
]. This is advantageous as a higher photon count decreases the Poisson noise in the camera system. A further advantage of ZnCd compared to Gd-based scintillation screens is the good
discrimination. Hence, the combination of fast buildup and decay, as well as high overall light yield with a high light yield stability makes ZnCd a good alternative to ZnS especially for high frame rate neutron imaging.
Comparing our result for the decay time of ZnS (
) to 1% to previously reported decay times of 85 μs to 10% [5
] there is a large discrepancy. While one explanation of the differences in decay time could be attributed to differences in the doped ZnS scintillation compound, a difference on the order of more than 1 × 105
seems unlikely. As we have noted, the decay of the light yield does not follow a single exponential decay, hence it may be that the ZnS screen we have measured also decayed to 10% in microseconds, which was however not accessible with the time resolution of our measurements. Another explanation for the different results may be down to differences in the excitation of the scintillation material. In [5
], it seems that the provided decay curve of a ZnS/6
LiF scintillation screen was measured using optical light with 337
wavelength to excite the scintillation screen. For other scintillation screen materials it has been shown previously that the type of exciting radiation influences the decay time [31
]. Hence, we expect also for ZnS/6
LiF that the measured decay time depends on the excitation radiation. As such, great care must be taken when comparing different scintillation screen decay time measurements, as the type and the intensity of the excitation radiation must be taken into account. Furthermore, as we have also seen in our experiments, thermal noise inherent to the detector system may hide light yield decay.
In conclusion, a Gd-based screen is best suited for high spatial resolution neutron imaging, while ZnCd is most suited for high frame rate neutron imaging. To further understand the mechanisms governing the response of neutron scintillation screens, especially ZnS, to varying flux levels, it is planned to perform buildup and decay measurements at different timescales at high flux sources. ZnCd in particular will remove limitations in neutron imaging frame rates, as the fast decay reduces the ghosting effect i.e., information of one frame in another. This will also allow the acquisition of more quantitative results as one systematic error in the data can be eliminated.