1 Introduction

Because of the persistence of the energy crisis, controllable fusion energy has drawn people’s attention. The agreement for the ITER (International Thermonuclear Experimental Reactor) project was signed between seven countries in 2006. Its purpose is the collaborative research of practical controllable fusion technology. Of the seven countries, China has taken responsibility for a series of detection projects.

Most of our attention concerns the operating conditions and parameters of the plasma in the ITER reactor. A series of plasma diagnostic systems have been proposed [1-3]. The NFM system with a fission chamber as the detector is an important diagnostic system within the ITER. Because the neutron flux is a significant parameter in the calculation of fusion reaction power, the spatial and time distribution of the plasma varies with discharge situations. For example the plasma ion temperature and density distribution can vary [4,5]. After considerable studies for scientific feasibility, two fission chambers were designed and manufactured for the NFM system. They serve as the neutron acquirement front ends [6-8].

In the ITER reactor, high temperature plasma radiates a large number of hard x rays, ɣ rays, and neutrons. Background radiation will severely influence the measurement results from the fission chamber. Therefore rigorous shielding must be applied around the fission chamber [9-14]. Research on FC shielding and moderators began with the design of the NFM system. In 2009 simulations used polyethylene and graphite as a moderator. The simulations mainly considered the intrinsic detecting efficiency over different incident angles [15-17]. In 2012 beryllium and boron carbide were considered as moderator materials[18]. However, there remain two deficiencies in this research. First, all discussion of the fission chamber moderator was made without confirmed FC parameters, especially the ɣ ray sensitivity. Therefore the suggested thickness of the moderator is generally higher than the actual requirement. Second, all the discussions about the moderator lacked the support of experimental results. Here we offer a new simulation using a low thickness beryllium moderator with a thickness between 20mm and 50mm. Furthermore, we conducted an experiment to measure the FC parameter and tested the effectiveness of the moderator. The results agree well with our simulated results.

The fusion reaction in ITER contains two stages: D-D fusion and D-T fusion. The neutron energy spectrum in the D-D process is a Gauss Fusion Spectrum, with an average energy reading of 2.45 MeV. In the D-T process, it is 14.1 MeV. That means the sensitivity of the FC in the D-D processes should be higher than that of the D-T process. Therefore, early work applied Uranium-235 (U-235) as the fission material with beryllium as the moderator in D-D process. And in D-T fusion, Uranium-238 (U-238) was applied as a fission material, with boron as a shield material. The combination of two FCs with different sensitivities is a good method for obtaining a wide range of neutron measurements. Also, the combination demonstrates that the moderators lead to a flat energy curve when the magnitude of beryllium moderator thickness is about 100mm; this also ensures the other particles likely are shielded [19].

In the experiment that measures the FC parameter, we find that the FC is actually extremely insensitive to ɣ rays. In the experiment the FC is set, without shielding, within a very high Cobalt-60 (Co-60) radiation field, with the absorbed dose at about 100 Gy/h. But the measured count is about 40cps. From the energy spectrum of Co-60 we can estimate that the ɣ ray detection efficiency of the FC has a magnitude of 10^-9. Since the detection efficiency of the FC we used in the experiment is 1 (unless the ɣ ray flux is extremely larger than the neutron flux), it may not be necessary to use multi-layer shielding, or very thick moderators, to eliminate the ɣ ray’s contribution. This experimental detail will be introduced with the moderator test in the Experiment section. Therefore, our simulation covers the low thickness situation, with the beryllium moderator between 20-mm and 50-mm. In the end we will provide the FC detection efficiency of each simulated situation.

2 Simulation

According to the analysis the magnitude of the energy spectrum in the ITER reaction ranges from 0.1 MeV to 10 MeV. As a result we chose seven energy levels, based on the energy spectrum of mono-energetic neutron sources within the experiment. The final selections are 0.1, 0.144, 0.25, 0.565, 0.75, 1, 1.2, 2.5, 5, 7.5, 10, 12.5 and 14.8 MeV. The processing of the FC is simulated with the Monte Carlo particle transport simulation tool Geant4. The simplified model is shown in Fig. 1a. The irradiated body contains three components: the stainless steel shell, the beryllium moderator, and the FC in the center. The thickness of the outermost shell is 3mm. The thicknesses for the simulation of the moderator are: 20, 30, 40, 42.5, 45, 47.5, and 50-mm. In the thickness range between 40-mm and 50-mm, we add three extra choices for simulation data to get the detailed results we desire within the vicinity of the experimental moderator thickness.

Fig. 1Full-screen View

Simplified model of FC and moderator (Top view) (a) and Structural schematic of FC used in ITER (b)

The FC model is simulated according to the manufactured FC designed for ITER. The structural detail of the ITER FC is shown in Fig. 1(b). According to the design handbook, it is a 50-mm × 500-mm cylindrical structure. The fissile layer is coated with 1-g of fissile material. Also, there is a 1-cm area, which is filled with argon, from the fissile layer to the electron collector. According to our analysis, we chose to use U-235 as the best option for fissile materials in D-D reaction, and U-238 in the D-T reaction. Since their structure is the same, this difference does not affect the simulation result.

The Geant4 simulation must be similar to the experimental environment. Fig. 2(b) is the schematic view of radiation scene as described by Geant4. In the figure we can see the neutron injection from the bottom of the figure, as well as the scattering in the moderator. At every energy dot we conduct a 10⁷ particles simulation. The results of the simulations are shown in Fig. 3.

Fig. 2Full-screen View

(Color online) The schematic view the radiation scene described by Geant4.

Fig. 3Full-screen View

(Color online) Simulation results of the detection efficiency with various thickness of the beryllium moderator

In Fig. 3 we can see the influence of a beryllium moderator. The x-coordinate refers to the various energy points. And the y –coordinate refers to the normalized detection efficiency (N/N_max). Here N_max refers to the maximum efficiency of the neutron detection obtained with all moderators. The curves going from the bottom to the top are the 20-mm to 50-mm beryllium moderators. It can be inferred that the detection efficiency will continue to increase as the thickness of the moderator grows larger than 50 mm in both the high energy and low energy segments.

3 Experiments

We provide two experiments to support the simulation results. First we need to ensure that the FC is not sensitive to other particles in ITER reactions especially those of ɣ ray reactions. So we conducted the ɣ ray radiation test in the China Institute of Measurement and Testing Technology. Second, we conduct the moderator effect experiment in the China Institute of Atomic Energy.

As we mentioned above, there are two FCs in the NFM system. One is manufactured by CNNC - Beijing Nuclear Instrument Factory (marked FC-I). The other is supported by ITER (marked FC-II). In the experiment we test the anti-ɣ and the moderator performance with both FCs.

3.1 The FC anti-ɣ performance Experiment

The FC anti-ɣ performance experiment is designed to ensure that the FC is not sensitive to other particles in D-D and D-T reactions. Since we can already discriminate for the ɑ particle, we mainly consider the influence of ɣ [20].

As shown in Fig. 4, part 1 is the FC. It is placed transversely at a distance from the radiation source such that the radiation adsorbed dose is about 100 Gy/h. Part 2 is the cobalt-60 (Co-60) source. It will be pulled out of the shield when the experiment begins. Part 3 is the connecting fiber to the NFM system. The FC count will be calculated during the end system.

Fig. 4Full-screen View

(Color online) Experimental environment of FC anti-ɣ performance

Because of the background counts, we must first measure the FC counts without radiation. Then we set the discrimination threshold to 0.32-V to eliminate the ɑ particles and record the data every 100 seconds. The results are listed in the following table, Table 1.

Table 1Full-screen View

The counting rate of FCs in the very high Co-60 radiation field

FC-I (with source) Counting Rate (cps)	FC-I (without source) Counting Rate(cps)	FC-II (with source) Counting Rate (cps)	FC-II (without source) Counting Rate (cps)
5443	4555	4464	4258
5326	4309	4436	4207
5321	4498	4560	4108
5308		4679
5285		4764
5406		4758
		4596
		4427
		4708

Show more

From Table 1 we can see that in such a high radiation field the average counts of FC-I and FC-II are 8.94-cps and 4.08-cps. From this experiment it can be ascertained that we are able to use the least thick moderator in simulation. In the beryllium moderator experiment we choose the 50mm beryllium layer as a moderator.

3.2 The beryllium moderator experiment

The experimental environment is shown in Fig. 5.

Fig. 5Full-screen View

(Color online) Environment of beryllium moderator experiments (a) and the comparative beryllium moderator experiments (b)

As shown in Fig. 5(a), part 1 is the moderator, which we used to make the simulation model in Fig. 1. The only difference is that the moderator is a 50mm thick beryllium ball layer. That leaves the equivalent thickness as an unknown. Parts 2 and 3 are the target and its accelerator, in which protons will be accelerated to hit the Li target. As a result, the accelerator will release neutrons. The accelerator is able to provide seven different mono-energetic neutrons. The provided energy levels are 0.144, 0.25, 0.565, 1.2, 2.5, 5, and 14.8 MeV. The energy spectrum examples of the accelerator are shown in Fig. 6 in which the mono-energetic neutron with 0.25 MeV is given. From this figure we can see the particle energy is concentrated in the small vicinity of 0.25 MeV. Therefore, it is able to meet our experimental requirements.

Fig. 6Full-screen View

The energy spectrum of typical mono-energetic neutron source

By considering the influence of the neutrons reflection off both the wall and the ground, we can conduct a comparative experiment to measure the influence of the reflected neutrons, in the same situation as shown in Fig. 5(b). In Fig. 5(b) part 1 is the neutron emission hole of the accelerator. Part 2 is a trapezoidal lead brick, which is able to fully shield the direct injection neutron. Part 3 is the FC, covered by a beryllium moderator, which is connected to the NFM system. In this experiment since the direct injection neutrons are almost all shielded by the lead brick, the NFM system counts mainly come from reflected neutrons.

At every energy point we measure the output 3~4 times for each FC, and calculate the average. After normalizing the results we arrive at the result shown in Fig. 7(a). For purpose of comparison we add the results of the 40-45 mm simulation to the figure. We also provide the original error analysis results in Fig. 7(b).

Fig. 7Full-screen View

(Color online) Results (a) and error analysis (b) of beryllium moderator experiments

In Fig. 7(b) the black and red lines are the result of the beryllium moderator experiment. The blue and green lines are the comparative experiment’s result. We can see that the standard deviation of each line is very small under the same experimental conditions. Since the whole experiment lasts for about 2 days, we can conclude that the experimental results are stable over a long period of time. Therefore, the results in Fig. 7(a) are accurate and credible.

The x-coordinate is the neutron energy, and the y-coordinate is the normalized detection efficiency. In the figure we can see that the measured results match the simulation well with the 45mm and 42.5mm curves. At the energy point of 14.8 MeV the relative error is a bit high. The relative errors between FC-I and FC-II, as well as the 42.5-mm curve for the energy point at 14.8 MeV are 14.67% and 4.97%, respectively. By considering the measuring error, we find that the FC-I result is acceptable and that the FC-II result matches the simulation fairly well.

4 Conclusion

The moderator of the ITER-NFM system is a problem that has been discussed since the beginning of the project. The simulations of the moderator material, its thickness and its incident angle have all been discussed in detail. For this study we selected beryllium as the moderator, and conducted a series of simulations and experiments. The greatest advantage of our work is the use of a low thickness moderator. It can help people save a lot of space and costs associated with solving other problems, e.g., FC electronic transmission, high voltage supply, and pre-amplifier shielding.

However, our work still has room for improvement. The moderator we used is a 50-mm layer filled with beryllium balls. It is very difficult to simulate the detailed situation without the diameter, distribution, and other parameters of the moderator. Therefore, our simulation is only approximately equivalent to the actual application in the ITER. Furthermore, the influence of the FC’s background radiation generated in ITER is not taken into consideration. Low strength moderation is based on the conclusion that there is no evidence to shows that the background radiation’s influence is much stronger than neutrons. We will continue to focus on this research area in the future.