Investigation of neutron imaging applications using fine-grained nuclear emulsion

Neutron imaging (NI) is a non-destructive technique for visualizing and analysing the inner structure of material objects. It has a wide range of applications Strobl and et al. (2009); Hussey and Jacobson (2015); Lehmann and et al. (2015); Kardjilov and et al. (2018); Tengattini et al. (2021); Song and et al. (2019); Lv and et al. (2018). The basic principle for NI is similar to that of X-ray imaging. Neutrons are neutral particles and can interact directly with atomic nuclei. The absorption and scattering cross-sections of neutrons depend on the nuclides. The dependence of attenuation coefficients of neutrons is different from that of X-rays which is correlated to the atomic number Banhart (2008). Contrary to X-rays, neutrons exhibit high attenuation for light (e.g. hydrogen, carbon, boron and lithium) and heavy (e.g. cadmium and gadolinium) nuclei. Lower attenuation is observed for heavy nuclei including aluminum, silicon, titanium and lead. These characteristics give neutron imaging an advantage over X-ray imaging. Neutron imaging may be employed as an effective method to visualize distributions for a wide range of elements using the difference in their neutron attenuation coefficients Strobl and et al. (2009).

The major challenges in developing NI techniques include improvements in spatial resolution Tremsin and et al. (2011); Hussey and et al. (2015); Trtik and et al. (2015); Trtik and Lehmann (2016); Bingham and et al. (2015); Morgano and et al. (2018); Trtik and et al. (2020), detection efficiency, Tremsin and et al. (2011) and time resolution Kardjilov and et al. (2018); Woracek and et al. (2019) of neutron detectors. Recent developments have improved the spatial resolution of NI to a few micrometers. One of the common approaches in NI is to convert the shading field of the neutron beam into an image of visible light using scintillators containing converters such as ⁶Li or ¹⁵⁷Gd Trtik and et al. (2020); Lehmann and et al. (2011); Hussey et al. (2017); Jiang and et al. (2021). The luminescence of a neutron capture event was observed by a charge–coupled device (CCD) or a complementary metal-oxide-semiconductor (CMOS) image sensor via an optical system. Using a technique based on this approach, the best resolution resolution currently reported is 2 $\mu$m Hussey et al. (2017). This spatial resolution was achieved using a thin gadolinium oxysulfide scintillator, magnifying optics, a CMOS image sensor, and an event positioning method by detecting the center of the luminescence for each event. Isegawa et al.Isegawa and et al. (2021) used a Gd₃Al₂Ga₃O₁₂:Ce single-crystal scintillator for the neutron radiography and achieved a spatial resolution of 10.5 $\mu$m.

For micrometer-scale spatial resolution or even better, nuclear emulsion is one of the potential candidates for NI device. Nuclear emulsion is a photographic film that can record the three-dimensional trajectories of charged particles with submicrometer-scale spatial resolution. It is composed of silver halide (AgBr) crystals dispersed in gelatin. Nuclear emulsions were initially developed for particle and nuclear physics experiments to observe the tracks of charged particles. They were used to observe pions Lattes, Occhialinidr, and Powelldr (1947), hypernuclei Danysz and Pniewski (1953); Davis (2005), charm particles Niu, Mikumo, and Maeda (1971) and double strangeness hypernuclei Danysz and et al. (1963). The analysis speed of nuclear emulsions has increased remarkably with improvements in image processing techniques Yoshimoto et al. (2017); Saito and et al. (2017). These improvements enable the use of nuclear emulsion in modern experiments, and led to the first direct detection of tau-neutrinos Kodama and et al. (2001), discovery of $\nu{{}_{\mu}}\rightarrow\nu{{}_{\tau}}$ oscillations in appearance mode Agafonova and et al. (2015), hypernuclei Saito and et al. (2017); Hiyama and Nakazawa (2018); Ekawa and et al. (2019); Hayakawa and et al. (2008); Yoshimoto and et al. (2021) and for several other applications including muon radiography Tanaka and et al. (2007); Morishima and et al. (2017).

Since the last decade, it has become possible to produce dedicated nuclear emulsions for specific purposes. Controlling the size of the AgBr crystals and chemical components contributed in achieving optimal spatial resolution and sensitivity. Fine-grained nuclear emulsions (FGNEs) with AgBr crystals, having a diameter of less than 50 nm, was developed Naka and et al. (2013); Asada et al. (2017) to detect the tracks of the nuclei recoiled by the so-called weakly interacting massive particles. The sensitivity of the nuclear emulsion was optimised to track the nuclei but was maintained sufficiently low for electrons and $\gamma$-rays Shiraishi and et al. (2021).

The FGNE with crystal diameter of 40 nm was employed for cold/ultracold neutron detection by combining it with a neutron converter layer formed by ¹⁰B₄C. This layer absorbs neutrons and emits two charged particles heading in the opposite direction via the following process: ¹⁰B + n $\rightarrow$ ⁷Li + ⁴He. When one of the emitted ions passes isotropically through an emulsion layer, latent image specks are created in some AgBr$\cdot$I crystals along its trajectory. After the chemical development, each crystal that contains latent image specks changes to an enlarged silver grain. A series of these grains is recognized as a track of the charged particle. Naganawa et al.Naganawa and et al. (2018) reported the development of a cold/ultracold neutron detector using FGNE and that the spatial resolution of neutron absorption points was estimated as less than 0.1 $\mu$m from linear fitting of tracks. The track lengths of ⁷Li and ⁴He recorded in the emulsion layer were 2.6 $\pm$ 0.4 and 5.1 $\pm$ 0.4 $\mu$m, respectively. This development was initiated for fundamental physics experiments to investigate quantized states under the influence of the earth’s gravitational field and to search unknown interactions attracting neutrons Nesvizhevsky and et al. (2002); Abele et al. (2009); Ichikawa and et al. (2014); Muto and et al. (2022).

The FGNEs combined with a boron-based neutron converter can be applied for NI applications. For the purpose of NI, the accumulated track density must be several orders of magnitude higher than that used in ref. Naganawa and et al. (2018). Hirota et al. Hirota and et al. (2021) demonstrated NI of a crystal oscillator chip using FGNE under the accumulated track density of $3\times 10^{4}$ tracks per 100 $\times$ 100 $\mu$m². They used a similar NI detector as in ref. Naganawa and et al. (2018), but the thickness of ¹⁰B₄C layer was 2 $\mu$m to increase neutron conversion efficiency. The thickness of ¹⁰B₄C layer used for neutron detection in ref. Naganawa and et al. (2018) was 50 nm. Gold wires with a diameter of approximately 30 $\mu$m in the crystal oscillator chip were successfully visualized. Furthermore, they attempted NI of a gadolinium-based grating with a periodic structure of 9 $\mu$m under the accumulated track density of $6\times 10^{3}$ tracks per 100 $\times$ 100 $\mu$m². They confirmed the grating spacing of 9 $\mu$m pitch using Rayleigh test with a standard deviation of 1.3 $\mu$m. They concluded that the achievable resolution for the NI using the FGNE was better than 3 $\mu$m. The challenges presented in ref. Hirota and et al. (2021) were to develop a method of image analysis and spatial resolution under a condition of $\sim$ $10^{4}$ tracks per 100 $\times$ 100 $\mu$m². With such a high track density, the conventional track-by-track analysis becomes difficult because of the overlap of the tracks.

For NI, the neutron detector based on the FGNE combined with neutron converter has several merits. It can detect neutron capture events with submicrometer-scale spatial resolution. The detector is lightweight, thin, small ( adjustable to sample objects with an area of approximately 100 cm²), and does not require any electric supply or drive in electronics. Furthermore, this detector can be integrated into devices or sample environment apparatus to shorten the distance with the investigated object. It can be added to existing imaging systems without any major modification and may be used as a complementary imaging device. Along with these merits, there are several limitations as well. A longer beam exposure time is required for a thinner ¹⁰B₄C layer in order to achieve higher spatial resolution. An additional process of chemical development is required after exposure. Furthermore, this detector is unsuitable for real-time measurements and applications where neutron energy measurement is required.

For NI applications using the FGNE, a high density accumulated tracks is a basic requirement. We describe our detector in section II. The estimation of track density and diffuseness as a metric to evaluate the imaging resolution is discussed in section III.1. The procedure employed to investigate the contrast resolution is discussed in section III.2. The details of the experiments are discussed in section IV. The results for NI of gadolinium-based gratings are discussed in section V. The NI application of the FGNE is discussed in the section VI. We finally conclude our findings and discuss future prospects in section VII.

The neutron detector developed for this work, shown in Figure 1a, has a layered structure. The base of the detector is a 0.4 mm thick silicon substrate. On this substrate, ¹⁰B₄C, NiC and C layers of thicknesses 230 nm, 46 nm and 14 nm, respectively, were formed by ion beam sputtering technique Hino and et al. (2015). The thickness of the ¹⁰B₄C layer was determined by taking into account neutron detection efficiency, the physical stability, and the spatial resolution for the imaging. The ¹⁰B₄C layer converts an incident neutron into ⁴He and ⁷Li. The NiC layer was used to physically stabilise the ¹⁰B₄C layer. The C layer was used for chemical protection from the NiC layer and for providing a strong adhesion to the emulsion. An emulsion layer of thickness 10 $\mu$m was formed on the sputtered layers in a dark room by pouring and drying the FGNE. It was composed of AgBr$\cdot$I crystals of approximately 40 nm diameter uniformly dispersed in medium including gelatin and polyvinyl alcohol. The neutron detector was packed with a light- and air-tight laminated bag, composed of nylon, polyethylene, aluminum, polyethylene and black polyethylene layers having thicknesses of 15 $\mu$m, 13 $\mu$m, 7 $\mu$m, 13 $\mu$m and 35 $\mu$m, respectively. During the beam exposure, the detector and the gadolinium-based grating were placed close together to minimise the blurring effects induced by the spreading of transmitted neutron beam, as shown in Figure 1a. Some incident neutrons were absorbed in the grating, and remaining neutrons were absorbed in the ¹⁰B₄C layer. When a ¹⁰B nucleus absorbs a neutron, the reaction products, ⁴He and ⁷Li, are emitted head off in opposite directions. When one of them passes through the emulsion layer, latent image specks are created in AgBr$\cdot$I crystals along its trajectory. After chemical development, crystals containing latent image specks transform to silver grains. Figure 1b shows a photograph of the detector after the chemical development. After chemical development, the thickness of the emulsion layer shrinks by a factor of 0.55 $\pm$ 0.1. A set of gray scale images were obtained with submicrometer precision using an epi-illumination optical microscope. The series of silver grains in these images is recognized as the induced tracks of the charged particles.

Tracks of ⁴He and ⁷Li have certain length as they travel through the nuclear emulsion. If the track density is high ($\sim$ $10^{5}$ tracks per 100 $\times$ 100 $\mu$m²), they blot out the edge of the object in the obtained images. On the other hand, if the track density is low ($\sim$ $10^{2}$ tracks per 100 $\times$ 100 $\mu$m²), the outline of the object will not be visible. Therefore, it was important to estimate the required track density for NI, and diffuseness as a function of track density for the current NI system. The track density required for NI, and diffuseness was estimated using simulated images which we present below.

For the neutron irradiation, we used the low-divergence beam branch Mishima and et al. (2009) of BL05 in the Materials and Life Science Experimental Facility (MLF) at the Japan Proton Accelerator Research Complex (J-PARC) Nakajima and et al. (2017). This is one of the most suitable beam branches in the MLF for this experiment because it provides a reasonably intense neutron beam owing to a coupled neutron moderator.

For NI of the gadolinium-based grating slit, the aperture of the slit was 44 mm and 6 mm for the horizontal and vertical directions, respectively. The beam divergence in the horizontal and vertical directions was set to 0.3 mrad and 10 mrad, respectively. It is to be noted that the beam divergence parallel to the grating does not affect the blurring of the image. The neutron flux was $2\times 10^{6}$ n/cm²/s, and the neutron irradiation time was 2.8 h to accumulate approximately $2\times 10^{4}$ tracks per (100 $\mu$m)². The distance between the gadolinium-based grating and the ¹⁰B₄C layer was 1.5 mm. The detector and the grating were fixed by metal clips and mounted on a holder for optical elements on an optical table.

On the other hand for imaging applications using the FGNE, NI of the Siemens star test pattern was also performed. The aperture of the slit on the 12 m position was set to 4 mm and 6 mm for the horizontal and vertical directions, respectively, to avoid blurring due to vertical divergence. The corresponding horizontal and vertical divergences of the beam were 0.3 mrad and 1.0 mrad, respectively. The typical beam power and the neutron flux during the experiment were 0.7 MW and $10^{5}$ n/cm²/s, respectively. The irradiation time was 9 h to accumulate $\sim$ $1\times 10^{4}$ tracks per (100 $\mu$m)². The distance between the gadolinium layer of the Siemens star and the ¹⁰B₄C layer in the detector was 1.1 mm. The NI experiments for the grating slit and the Siemens star test patter were carried out under a stable environmental temperature of 23.9 $\pm$ 0.1 ^∘C.

As discussed in the section IV, NI of the Siemens star test pattern was performed for the demonstration one of the applications of the proposed NI system. Figure 18a shows a photograph of the Siemens star test pattern used in our experiments. This pattern was designed by a group in the Paul Scherrer Institute (PSI), Villigen, Switzerland. A prototype of the test pattern is described in Ref. Grünzweig and et al. (2007). The pattern consisted of 128 spokes in a circular area of 20 mm diameter and was made of a thin layer ($\sim$ 5 $\mu$m) of gadolinium on a quartz wafer of thickness 0.7 mm. The space between the spokes became narrower towards the center of the circle. The innermost spoke had a period of 10 $\mu$m, which corresponds to 100 line pairs per millimeter. Figure 18b shows an optical image of the Siemens star test pattern using optical microscope with a trans-illumination system with a magnification of 20x, and a numerical aperture (NA) of 0.35.

The spatial resolution of an imaging device is assessed by the fine resolution of the intervals of the inner spokes. Figure 18b shows a photograph of the developed detector irradiated with the neutron beam through the pattern. The pattern transferred on the emulsion layer is visible to the naked eye even before being observed under an optical microscope. The dark regions correspond to the sites where silver grains were developed by the emitted ⁴He or ⁷Li during the neutron absorption events. Other regions correspond to sites where the neutrons were absorbed by gadolinium on the pattern. The neutron absorption events in the ¹⁰B₄C layer were suppressed. Figure 18c shows an image of the nuclear emulsion, which recorded the thinnest spokes near the center of the pattern. As shown in Figure 18b, the grating is not very well fabricated therefore the diffuseness for the edges of the grating was not evaluated quantitatively. One can note that the innermost spokes are clearly resolved as shown in Figure 18d. Our NI results shows the ability of the FGNE to resolve the micrometer-scale structures with micrometer-scale resolution or even better.

In this work, we presented the simulated results of neutron imaging using the fine-grained nuclear emulsion. It was concluded that a track density of $\sim$ $10^{4}$ tracks per 100 $\times$ 100 $\mu$m² was required for neutron imaging applications. The contrast resolution was investigated for various track densities under varying conditions of neutron transmission rates. Our results estimated a contrast resolution and area capable of separation 1$\sigma$ per ($\mu$m²) under various accumulated track densities per (100 $\mu$m)² and varying neutron transmission rates for the current neutron imaging system.

The neutron imaging of gadolinium-based gratings with known geometries, such as a Siemens star test pattern and a grating slit with a periodic structure of 9 $\mu$m, were further reported in this work. These micrometer structured gratings were imaged and resolved successfully. The evaluated diffuseness for the edges of the grating slit under high accumulated track density conditions, for simulated and optical images, were 0.863 $\pm$ 0.004 $\mu$m and 0.945 $\pm$ 0.004 $\mu$m, respectively. The difference in the diffuseness values may be attributed to the shape of the gadolinium teeth. It is reminded that the grating slit is not a perfect rectangular structure because of the production process, as described in Ref. Samoto, Takano, and Momose (2019a, b). The diffuseness values for simulated and optical images were in fair agreement and it may be concluded that the achievable resolution under the accumulated track density of $3\times 10^{4}$ tracks per 100 $\times$ 100 $\mu$m² is 0.945 $\pm$ 0.004 $\mu$m. The obtained resolution for the optical images is an inclusive assessment that includes the effects of the resolution of the detector and the shape of the gadolinium tooth. It is concluded that the actual resolution may be better than the deduced resolution, and neutron imaging with submicrometer spatial resolution can be performed using the neutron detection system based on the fine-grained nuclear emulsion. In future, we plan to perform experiments for the exclusive assessment of simpler structure and evaluate the spatial resolution of the proposed neutron imaging system including neutron imaging of material objects.

The data that support the findings of this work are available from the corresponding author upon reasonable request.