- Email: [email protected]
Design of 500 MHz RFQ linear accelerator for a compact neutron source, RANS-III
Nuclear Inst. and Methods in Physics Research B 461 (2019) 186–190
Contents lists available at ScienceDirect
Nuclear Inst. and Methods in Physics Research B journal homepage: www.elsevier.com/locate/nimb
Design of 500 MHz RFQ linear accelerator for a compact neutron source, RANS-III
T
Shota Ikedaa, , Yoshie Otakea, Tomohiro Kobayashia, Noriyosu Hayashizakib ⁎
a b
Riken, 2-1, Hirosawa, Wako, Saitama 351-0198, Japan Tokyo Institute of Technology, 2-12-1 Ookayama, Meguro-ku, Tokyo 152-8550, Japan
ARTICLE INFO
ABSTRACT
Keywords: RFQ linac Proton accelerator Compact neutron source
A compact neutron source using a particle accelerator is a promising tool for material analysis, infrastructural diagnostics, fissile material detection, and medical applications. For non-destructive, on-site inspection of the degradation of aged concrete and steel reinforcement it is essential to develop a light and compact neutron source system. To design a more compact RFQ linac for a transportable neutron source, we developed a 500 MHz RFQ linac. In this paper the design of a 500 MHz RFQ linac for a compact neutron source is described. First, using beam tracking simulation software, we designed the cell parameters of the 500 MHz RFQ linac, which can accelerate a proton beam from 30 keV to 2.49 MeV. Then, using three-dimensional electromagnetic simulation software and multiphysics simulation software, we optimized the cavity design of the 500 MHz RFQ linac, and investigated the thermal properties of the cavity with the cooling channels.
1. Introduction With the aging and decommissioning of nuclear reactors, accelerator-driven neutron sources are being developed around the world. Recently, accelerator-driven compact neutron sources, which can be installed and operated for the evaluation of materials in laboratory units, have become increasingly desirable. There are two types of accelerator-driven compact neutron sources, namely, a proton linac with beryllium or lithium target, or an electron linac with heavy-metal target. In particular, at RIKEN, the RIKEN accelerator-driven compact neutron source (RANS) [1] began operation in 2013 for the non-destructive evaluation of materials used in industry and infrastructure (Fig. 1(a)). With the progress of neutron diffractometer methods and quantitative evaluation methods for infrastructure imaging, we have successfully established the technological development that can perform quantitative evaluation and analysis even with a low-yield neutron beam (1012 neutrons/s). A more compact neutron source (RANS-II) [2] for on-site use is being developed with a 2.49 MeV, an average beam current of 100 μA, 200 MHz proton RFQ linac, which has three main components (a major vane and two minor vanes) and a lithium target. (Fig. 1(b)) The total neutron yield for an average proton current of 100 μA is estimated to be ⁎
approximately 1011 neutrons/s and the neutron emission is dominant in the forward direction, which is estimated to be sufficient to distinguish the voids in thick concrete slabs using the fast neutron imaging technique. To realize a novel transportable compact neutron source, RANS-III, we are now developing a 500 MHz RFQ linear accelerator. The comparison of the linear accelerators is summarized in Table 1. In the 500 MHz RFQ, the cavity diameter is 1.8 times smaller and the weight is ~66% lighter than that in a RFQ at more often used frequency (lower than 450 MHz), because the resonance frequency is inversely proportional to the cavity diameter. Because the RANS-III system needs to generate a neutron yield as high as that of RANS-II using a lithium target, the 500 MHz RFQ needs to accelerate the protons to 2.49 MeV with an average beam current of 100 μA at a duty factor of 3%. As part of the development, we designed and evaluated a 500 MHz RFQ, using several simulation software (beam tracking, electromagnetic, and thermal simulation). 2. Beam dynamics simulation We simulated the output beam current for the length of the RFQ electrode using the RFQ design code, LIDOSRFQ [3]. In a three-component RFQ linear accelerator, it is difficult to restrict the minimum
Corresponding author. E-mail address: [email protected] (S. Ikeda).
https://doi.org/10.1016/j.nimb.2019.09.051 Received 10 June 2019; Received in revised form 25 September 2019; Accepted 26 September 2019 0168-583X/ © 2019 Elsevier B.V. All rights reserved.
Nuclear Inst. and Methods in Physics Research B 461 (2019) 186–190
S. Ikeda, et al.
Fig. 1. Riken accelerator-driven compact neutron source. Table 1 Accelerator parameters of RANS, RANS-II, and RANS-III.
Target Accelerator type Frequency (MHz) Output beam energy (MeV) RF peak power (kW) Duty (%) Peak bean current (mA) Average beam current (µA) Beam power (W) Accelerator length (cm) Accelerator cross section (cm2) Accelerator weight (t)
RANS
RANS-Ⅱ
RANS-III
Be RFQ + DTL 425 7 550 1.3 ≦10 100 700 450 1850 ≃5
Li RFQ 200 2.49 ≦200 3.3 ≦10 ≧100 ≧250 350 2000 ≃3
Li RFQ 500 2.49 ≦250 3.0 ≦10 ≧100 ≧250 280 620 ≃1
aperture radius to less than 1.5 mm owing to structural limitations. In an RFQ, the focusing strength (B) is proportional to: V / af 2 with f the resonant frequency, a the average beam aperture and V the vane voltage. The focusing strength of a 500 MHz RFQ tends to be low. Therefore, it is difficult to design cell parameters that can accelerate a several mA proton beam. Simulations are performed in LIDOSRFQ in a two-step process: 1. Matrix calculation for designing the cell parameters and 2. Particle trajectory analysis by particle in simulation (PIC) using electric field distribution at the designed RFQ electrode. The cell parameters of the 500 MHz linac are illustrated in Fig. 2, with a minimum aperture radius, m the modulation index, W the beam energy, Phi the synchronous phase, and r0 the average aperture radius. The RFQ electrode length is ~2230 mm, two-thirds of which constitutes the accelerator section. The minimum aperture radius is 1.75 mm,
Fig. 3. Simulated output beam profile (Input emittance: 0.2 π mm mrad).
Fig. 2. Cell parameters of the 500 MHz linac. 187
Nuclear Inst. and Methods in Physics Research B 461 (2019) 186–190
S. Ikeda, et al.
The simulated output beam profiles and the beam parameters are illustrated and presented in Fig. 3 and Table 2. A 12 mA proton beam, which is a peak beam current of an ECR ion-source under development [4], with a transverse 4D waterbag, an input energy of 30 kV, and a transverse normalized envelop emittance of 0.2 π mm mrad was used for the PIC simulation. The simulation resulted in an accelerated beam of 9.02 mA and an output energy of 2.49 MeV. Because the fringe field section is provided near the exit of the RFQ electrode, the output beam is in a gently diverging state both horizontally and vertically. We calculated the beam transmission and output peak beam current as a function of the input beam emittance (Fig. 4), both of which tend to decrease with the increase in emittance. When the accelerator duty factor is 3%, the peak beam current must be 3.3 mA or more to obtain an average beam current of 100 μA, such that the total transverse normalized emittance of the beam incident on the RFQ should be less than 2.0 π mm mrad (total normalized).
Table 2 Calculated beam parameters of the RANS-III RFQ. Particle species
Proton
Electric field strength (kilp) Vane voltage (kV) Input beam peak current (mA) Input beam energy (keV) Input emittance (mm mrad, 6RMS) Transmission (%) Output peak beam current (mA) Output beam energy (MeV) Average aperture radius (mm) Minimum aperture radius (mm) RFQ electrode length (mm) Cavity length (mm)
1.35 58.5 12 30 0.2 75.6 9.02 2.49 2.5 1.71 2,232.5 2,372.5
3. Cavity design 3.1. Electromagnetic simulation We optimized a cross section of the cavity of the 500 MHz RFQ linac using three-dimensional electromagnetic simulation software. The optimizer, which is based on the Nelder–Mead method, was used to obtain the cavity cross-sectional shape that minimizes the wall loss at the resonant frequency of 500 MHz as a function of the cavity parameters, as illustrated in Fig. 5. In this optimization, from other design result, the cavity length was 2350 mm and the average aperture radius was 2.5 mm. The optimized cavity parameters are listed in Table 3. The resonance frequency was 500 Hz and the wall loss normalized with respect to the inter-rod voltage was 183.3 kW. Owing to the influence of the modulation of the RFQ electrode, the quadrupole field strength distribution changes. In the RFQ linear accelerator for RANS-III, because the inner diameter of the cavity is smaller than the cavity length, it is difficult to ensure uniform quadrupole electric field strength only by the adjustment of the end cut. In addition, we examined the adjustment of the electromagnetic field intensity distribution by tuner blocks (Fig. 6). To ensure that the quadrupole field strength distribution is uniform along the beam axis, the thickness of the tuner block in the acceleration cavity was adjusted using the three-dimensional electromagnetic simulation software. The quadrupole field strength distribution before and after the tuning of the tuner blocks is illustrated in Fig. 7. After the optimization, the end cut had an incident side of 44 mm, an exit side of 41 mm, a thickness of 1.85 mm for the tuner blocks 1 and 2, and a thickness of 0.6 mm for the tuner block 3. By adjusting the end-cut depth and the block-tuner thickness, the inter-vane voltage was reduced from 80% to 14%.
Fig. 4. Transmission vs input beam emittance.
Fig. 5. Optimized cavity cross-section.
3.2. Thermal simulation
Table 3 Optimized cavity parameters. a (mm) b (mm) c (mm) d (mm) e (mm) f (mm) Resonance frequency (MHz) Unloaded Q Wall loss (100% Q) (kW)
The wall loss causes an increase in the temperature of the cavity, and the shape of the electrode is deformed, which adversely affects the RF property. To reduce the temperature increase, cooling water is circulated in 6 mm diameter vane channels and body channels on the outside of the RFQ, as illustrated in Fig. 8. Under the conditions of 244.4 kW (75% Q) wall loss and 3% duty factor, the temperature increase due to the RF power loss and the resonance frequency shift owing to the cavity deformation were simulated using three dimensional multiphysics simulation software CST MPHYCHCS STUDIO [5]. The simulation conditions are listed in Table 4. The temperature peak was located along the edges of the end cut, and increased from 20.0 to 28.1 °C. From the simulation results, we
5.89 8.47 65.8 70.0 39.8 54.0 500.3 8736 183.3
average radius is 2.5 mm, inter-vane voltage is 58.5 kV, and maximum electric field strength is 31.5 MV/m, which is 1.35 times the Kilpatrick factor. 188
Nuclear Inst. and Methods in Physics Research B 461 (2019) 186–190
S. Ikeda, et al.
Fig. 6. Tuner blocks. Table 4 Conditions of thermal simulation. Wall loss (75% Q) (kW) Duty factor (%) Cooling channels Cross section (Vane ch) (mm2) Cross section (Body ch) (mm2) Room temperature (°C) Water temperature (°C) Cavity material
Fig. 7. Quadrupole field distribution.
244.4 3 Vane and body 28 16 25 20 Copper
electromagnetic simulation software, beam tracking simulation software, and thermal simulation software, and optimized the cavity design and cooling channel that can accelerate a proton beam from 30 keV to 2.49 MeV with a duty cycle of 3%. We have calculated the minimum emittance in order to obtain an average beam current of 100 μA, and the total transverse normalized emittance of the input beam with 12 mA and a duty cycle of 3% should be less than 2.0 π mm mrad. Based on the design result, a prototype cavity will be fabricated and a beam acceleration test with an ECR ion source and a solid-state RF amplifier (R&K-CA114BW2-6171RP [6]) will be conducted in the future.
estimate a resonance frequency shift of less than 19 kHz, and therefore conclude the 500 MHz RFQ linac is able to operate under the conditions of 244.4 kW wall loss and 3% duty factor. 4. Summary and future plan For a novel transportable compact neutron source, RANS-III, we designed a 500 MHz RFQ linac using three-dimensional
Fig. 8. Cooling channel layout.
189
Nuclear Inst. and Methods in Physics Research B 461 (2019) 186–190
S. Ikeda, et al.
References
[3] < http://www.ghga.com/accelsoft/lidosrfq.html > . [4] N. Hayashizaki et al, Development of a downsized proton accelerator system for compact neutron sources, Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms, in press. [5] < http://www.cst.com > . [6] < http://www.rkmicrowave.com/products/pdf/CA500BW2-8085RP.pdf > .
[1] Y. Otake, A compact proton linac neutron source at RIKEN, Applications of Laserdriven Particle Acceleration, CRC Press, 2018, pp. 291–314. [2] T. Kobayashi, et al., Development of accelerator-driven transportable neutron source for non-destructive inspection of concrete construct, 2017 IEEE Nuclear Science Symposium and Medical Imaging Conference (NSS/MIC), IEEE, 2017, pp. 1–2.
190
Contents lists available at ScienceDirect
Nuclear Inst. and Methods in Physics Research B journal homepage: www.elsevier.com/locate/nimb
Design of 500 MHz RFQ linear accelerator for a compact neutron source, RANS-III
T
Shota Ikedaa, , Yoshie Otakea, Tomohiro Kobayashia, Noriyosu Hayashizakib ⁎
a b
Riken, 2-1, Hirosawa, Wako, Saitama 351-0198, Japan Tokyo Institute of Technology, 2-12-1 Ookayama, Meguro-ku, Tokyo 152-8550, Japan
ARTICLE INFO
ABSTRACT
Keywords: RFQ linac Proton accelerator Compact neutron source
A compact neutron source using a particle accelerator is a promising tool for material analysis, infrastructural diagnostics, fissile material detection, and medical applications. For non-destructive, on-site inspection of the degradation of aged concrete and steel reinforcement it is essential to develop a light and compact neutron source system. To design a more compact RFQ linac for a transportable neutron source, we developed a 500 MHz RFQ linac. In this paper the design of a 500 MHz RFQ linac for a compact neutron source is described. First, using beam tracking simulation software, we designed the cell parameters of the 500 MHz RFQ linac, which can accelerate a proton beam from 30 keV to 2.49 MeV. Then, using three-dimensional electromagnetic simulation software and multiphysics simulation software, we optimized the cavity design of the 500 MHz RFQ linac, and investigated the thermal properties of the cavity with the cooling channels.
1. Introduction With the aging and decommissioning of nuclear reactors, accelerator-driven neutron sources are being developed around the world. Recently, accelerator-driven compact neutron sources, which can be installed and operated for the evaluation of materials in laboratory units, have become increasingly desirable. There are two types of accelerator-driven compact neutron sources, namely, a proton linac with beryllium or lithium target, or an electron linac with heavy-metal target. In particular, at RIKEN, the RIKEN accelerator-driven compact neutron source (RANS) [1] began operation in 2013 for the non-destructive evaluation of materials used in industry and infrastructure (Fig. 1(a)). With the progress of neutron diffractometer methods and quantitative evaluation methods for infrastructure imaging, we have successfully established the technological development that can perform quantitative evaluation and analysis even with a low-yield neutron beam (1012 neutrons/s). A more compact neutron source (RANS-II) [2] for on-site use is being developed with a 2.49 MeV, an average beam current of 100 μA, 200 MHz proton RFQ linac, which has three main components (a major vane and two minor vanes) and a lithium target. (Fig. 1(b)) The total neutron yield for an average proton current of 100 μA is estimated to be ⁎
approximately 1011 neutrons/s and the neutron emission is dominant in the forward direction, which is estimated to be sufficient to distinguish the voids in thick concrete slabs using the fast neutron imaging technique. To realize a novel transportable compact neutron source, RANS-III, we are now developing a 500 MHz RFQ linear accelerator. The comparison of the linear accelerators is summarized in Table 1. In the 500 MHz RFQ, the cavity diameter is 1.8 times smaller and the weight is ~66% lighter than that in a RFQ at more often used frequency (lower than 450 MHz), because the resonance frequency is inversely proportional to the cavity diameter. Because the RANS-III system needs to generate a neutron yield as high as that of RANS-II using a lithium target, the 500 MHz RFQ needs to accelerate the protons to 2.49 MeV with an average beam current of 100 μA at a duty factor of 3%. As part of the development, we designed and evaluated a 500 MHz RFQ, using several simulation software (beam tracking, electromagnetic, and thermal simulation). 2. Beam dynamics simulation We simulated the output beam current for the length of the RFQ electrode using the RFQ design code, LIDOSRFQ [3]. In a three-component RFQ linear accelerator, it is difficult to restrict the minimum
Corresponding author. E-mail address: [email protected] (S. Ikeda).
https://doi.org/10.1016/j.nimb.2019.09.051 Received 10 June 2019; Received in revised form 25 September 2019; Accepted 26 September 2019 0168-583X/ © 2019 Elsevier B.V. All rights reserved.
Nuclear Inst. and Methods in Physics Research B 461 (2019) 186–190
S. Ikeda, et al.
Fig. 1. Riken accelerator-driven compact neutron source. Table 1 Accelerator parameters of RANS, RANS-II, and RANS-III.
Target Accelerator type Frequency (MHz) Output beam energy (MeV) RF peak power (kW) Duty (%) Peak bean current (mA) Average beam current (µA) Beam power (W) Accelerator length (cm) Accelerator cross section (cm2) Accelerator weight (t)
RANS
RANS-Ⅱ
RANS-III
Be RFQ + DTL 425 7 550 1.3 ≦10 100 700 450 1850 ≃5
Li RFQ 200 2.49 ≦200 3.3 ≦10 ≧100 ≧250 350 2000 ≃3
Li RFQ 500 2.49 ≦250 3.0 ≦10 ≧100 ≧250 280 620 ≃1
aperture radius to less than 1.5 mm owing to structural limitations. In an RFQ, the focusing strength (B) is proportional to: V / af 2 with f the resonant frequency, a the average beam aperture and V the vane voltage. The focusing strength of a 500 MHz RFQ tends to be low. Therefore, it is difficult to design cell parameters that can accelerate a several mA proton beam. Simulations are performed in LIDOSRFQ in a two-step process: 1. Matrix calculation for designing the cell parameters and 2. Particle trajectory analysis by particle in simulation (PIC) using electric field distribution at the designed RFQ electrode. The cell parameters of the 500 MHz linac are illustrated in Fig. 2, with a minimum aperture radius, m the modulation index, W the beam energy, Phi the synchronous phase, and r0 the average aperture radius. The RFQ electrode length is ~2230 mm, two-thirds of which constitutes the accelerator section. The minimum aperture radius is 1.75 mm,
Fig. 3. Simulated output beam profile (Input emittance: 0.2 π mm mrad).
Fig. 2. Cell parameters of the 500 MHz linac. 187
Nuclear Inst. and Methods in Physics Research B 461 (2019) 186–190
S. Ikeda, et al.
The simulated output beam profiles and the beam parameters are illustrated and presented in Fig. 3 and Table 2. A 12 mA proton beam, which is a peak beam current of an ECR ion-source under development [4], with a transverse 4D waterbag, an input energy of 30 kV, and a transverse normalized envelop emittance of 0.2 π mm mrad was used for the PIC simulation. The simulation resulted in an accelerated beam of 9.02 mA and an output energy of 2.49 MeV. Because the fringe field section is provided near the exit of the RFQ electrode, the output beam is in a gently diverging state both horizontally and vertically. We calculated the beam transmission and output peak beam current as a function of the input beam emittance (Fig. 4), both of which tend to decrease with the increase in emittance. When the accelerator duty factor is 3%, the peak beam current must be 3.3 mA or more to obtain an average beam current of 100 μA, such that the total transverse normalized emittance of the beam incident on the RFQ should be less than 2.0 π mm mrad (total normalized).
Table 2 Calculated beam parameters of the RANS-III RFQ. Particle species
Proton
Electric field strength (kilp) Vane voltage (kV) Input beam peak current (mA) Input beam energy (keV) Input emittance (mm mrad, 6RMS) Transmission (%) Output peak beam current (mA) Output beam energy (MeV) Average aperture radius (mm) Minimum aperture radius (mm) RFQ electrode length (mm) Cavity length (mm)
1.35 58.5 12 30 0.2 75.6 9.02 2.49 2.5 1.71 2,232.5 2,372.5
3. Cavity design 3.1. Electromagnetic simulation We optimized a cross section of the cavity of the 500 MHz RFQ linac using three-dimensional electromagnetic simulation software. The optimizer, which is based on the Nelder–Mead method, was used to obtain the cavity cross-sectional shape that minimizes the wall loss at the resonant frequency of 500 MHz as a function of the cavity parameters, as illustrated in Fig. 5. In this optimization, from other design result, the cavity length was 2350 mm and the average aperture radius was 2.5 mm. The optimized cavity parameters are listed in Table 3. The resonance frequency was 500 Hz and the wall loss normalized with respect to the inter-rod voltage was 183.3 kW. Owing to the influence of the modulation of the RFQ electrode, the quadrupole field strength distribution changes. In the RFQ linear accelerator for RANS-III, because the inner diameter of the cavity is smaller than the cavity length, it is difficult to ensure uniform quadrupole electric field strength only by the adjustment of the end cut. In addition, we examined the adjustment of the electromagnetic field intensity distribution by tuner blocks (Fig. 6). To ensure that the quadrupole field strength distribution is uniform along the beam axis, the thickness of the tuner block in the acceleration cavity was adjusted using the three-dimensional electromagnetic simulation software. The quadrupole field strength distribution before and after the tuning of the tuner blocks is illustrated in Fig. 7. After the optimization, the end cut had an incident side of 44 mm, an exit side of 41 mm, a thickness of 1.85 mm for the tuner blocks 1 and 2, and a thickness of 0.6 mm for the tuner block 3. By adjusting the end-cut depth and the block-tuner thickness, the inter-vane voltage was reduced from 80% to 14%.
Fig. 4. Transmission vs input beam emittance.
Fig. 5. Optimized cavity cross-section.
3.2. Thermal simulation
Table 3 Optimized cavity parameters. a (mm) b (mm) c (mm) d (mm) e (mm) f (mm) Resonance frequency (MHz) Unloaded Q Wall loss (100% Q) (kW)
The wall loss causes an increase in the temperature of the cavity, and the shape of the electrode is deformed, which adversely affects the RF property. To reduce the temperature increase, cooling water is circulated in 6 mm diameter vane channels and body channels on the outside of the RFQ, as illustrated in Fig. 8. Under the conditions of 244.4 kW (75% Q) wall loss and 3% duty factor, the temperature increase due to the RF power loss and the resonance frequency shift owing to the cavity deformation were simulated using three dimensional multiphysics simulation software CST MPHYCHCS STUDIO [5]. The simulation conditions are listed in Table 4. The temperature peak was located along the edges of the end cut, and increased from 20.0 to 28.1 °C. From the simulation results, we
5.89 8.47 65.8 70.0 39.8 54.0 500.3 8736 183.3
average radius is 2.5 mm, inter-vane voltage is 58.5 kV, and maximum electric field strength is 31.5 MV/m, which is 1.35 times the Kilpatrick factor. 188
Nuclear Inst. and Methods in Physics Research B 461 (2019) 186–190
S. Ikeda, et al.
Fig. 6. Tuner blocks. Table 4 Conditions of thermal simulation. Wall loss (75% Q) (kW) Duty factor (%) Cooling channels Cross section (Vane ch) (mm2) Cross section (Body ch) (mm2) Room temperature (°C) Water temperature (°C) Cavity material
Fig. 7. Quadrupole field distribution.
244.4 3 Vane and body 28 16 25 20 Copper
electromagnetic simulation software, beam tracking simulation software, and thermal simulation software, and optimized the cavity design and cooling channel that can accelerate a proton beam from 30 keV to 2.49 MeV with a duty cycle of 3%. We have calculated the minimum emittance in order to obtain an average beam current of 100 μA, and the total transverse normalized emittance of the input beam with 12 mA and a duty cycle of 3% should be less than 2.0 π mm mrad. Based on the design result, a prototype cavity will be fabricated and a beam acceleration test with an ECR ion source and a solid-state RF amplifier (R&K-CA114BW2-6171RP [6]) will be conducted in the future.
estimate a resonance frequency shift of less than 19 kHz, and therefore conclude the 500 MHz RFQ linac is able to operate under the conditions of 244.4 kW wall loss and 3% duty factor. 4. Summary and future plan For a novel transportable compact neutron source, RANS-III, we designed a 500 MHz RFQ linac using three-dimensional
Fig. 8. Cooling channel layout.
189
Nuclear Inst. and Methods in Physics Research B 461 (2019) 186–190
S. Ikeda, et al.
References
[3] < http://www.ghga.com/accelsoft/lidosrfq.html > . [4] N. Hayashizaki et al, Development of a downsized proton accelerator system for compact neutron sources, Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms, in press. [5] < http://www.cst.com > . [6] < http://www.rkmicrowave.com/products/pdf/CA500BW2-8085RP.pdf > .
[1] Y. Otake, A compact proton linac neutron source at RIKEN, Applications of Laserdriven Particle Acceleration, CRC Press, 2018, pp. 291–314. [2] T. Kobayashi, et al., Development of accelerator-driven transportable neutron source for non-destructive inspection of concrete construct, 2017 IEEE Nuclear Science Symposium and Medical Imaging Conference (NSS/MIC), IEEE, 2017, pp. 1–2.
190