Introduction

Metallic components of spent nuclear fuel assembly, including Zircaloy hulls and stainless steel endpieces, are disposed in deep underground repositories after fuel reprocessing, owing to strong activation and contamination [1, 2]. Several studies have investigated 14C contents and their release from hulls [3,4,5,6,7,8] and activated steels [9,10,11,12,13]; additionally, speciation of 14C released from activated metals has also been analyzed [14,15,16]. Regarding the current post-closure safety case [1, 2], corrosion-related congruent release is a source of activated metallic wastes wherein the radionuclides are considered to be distributed homogeneously in the metal matrix. Therefore, radionuclide distribution is a key factor affecting the radionuclide release behavior, especially for the endpieces. This is because the neutron flux varies considerably at the edges of fuels, unlike the hull in the fuel section.

In this first experimental study on the irradiated endpiece, the content and distribution of 60Co, a typical activation product, in an endpiece were investigated prior to other radionuclides, such as 14C and 36Cl. The irradiated BWR lower tie plate was selected as the endpiece for analysis. The imaging plate (IP) method was applied because it can easily obtain comprehensive information on the radionuclide distribution over a relatively wide range. Moreover, the partial destructive analyses provided and verified the acquired IP data corrections. Activation calculations were also performed using a Monte Carlo burnup calculation code to be modeled based on neutron distribution and spectrum changes through the tie plate. The content of cobalt (59Co) impurities in the irradiated tie plate served as an input for accurate activation calculations.

Methodology

An irradiated lower tie plate made of stainless steel was obtained from the spent BWR assembly (Bundle ID: 2F1Z3) [17]. It represented a STEP III type fuel, a BWR fuel commonly used in Japan that uses a 9 × 9 array in a lattice configuration having two water rods. The fuels were loaded in the core of Fukushima Daini Nuclear Power Station Unit 1, and irradiation was conducted from July 1996 to May 2000 (for 1160 d in total) for three cycles. The average enrichment and burnup of assembly were 3.4% and 35.0 GWd/tHM, respectively. After the post-irradiation tests [17], the tie plate was stored with other structural materials for approximately 20 years in the storage pool and then supplied for this study.

The sample locations prepared for IP and ɤ-spectrometry are shown in Fig. 1. The specimen was thinned to approximately 1.0 mm by polishing to make it appropriate for IP measurements, in which all photostimulated luminescence (PSL) signals were assumed to be excited by 60Co because other short-lived ɤ-emitting nuclides such as 54Mn and 125Sb had almost decayed in the 20 year cooling, as shown in the ɤ-spectrum given in Fig. S1 where there are only ɤ-rays from 60Co including their single and double-escape peaks. While long-lived pure β-emitters, 14C, 36Cl, and 63Ni, were still present, they were small amount or low β-energy, and had little effect on the IP measurements. Due to the limit of air dose rate in the laboratory, the specimen was divided into three parts as shown in Fig. 1B–D. The Storage Phosphor Screen, BAS-MS (Cytiva), was used as an IP film. Based on the preliminary test results, the exposure time was set to 5 min for each sample. After the exposure, PSL was detected using the image analyzer Typhoon FLA7000 (GE Healthcare). We detected PSL multiple times, and the results of the 20th detection were used for image analysis, which was performed using the numerical calculation library (NumPy 1.18) and image processing library (OpenCV 4.5.5.62) in the programming language Python 3.8.10. A smoothing process was also performed for PSL signals. Subsequently, the effects of self-adsorption of radiation from 60Co by the specimen shape were corrected. The filter functions estimating the two-dimensional (2D) ɤ-ray energy transfer to IP were obtained using the Monte Carlo particle and heavy ion transport code system (PHITS Ver. 3.24) [18]. By applying the inverse function of the abovementioned filter function with 2D discrete Fourier transform, the distribution of radioactivity was re-constructed. Figure S2 presents the 2D model for self-adsorption correction using the PHITS simulation.

Fig. 1
figure 1

Left image represents the sampling positions for IP (A–D), ɤ-spectrometry (G1–G5), and cobalt impurity measurement (Co) in the lower tie plate. Middle and right images are the IP results

Full size image

The activation calculations for the lower tie plate were performed using the Monte Carlo burnup calculation system (MCNP-BURN2) [19], which constructs the object 3D structures and configures the nuclear cross sections of the components by implementing the activation function of ORIGEN2 code [20]. In the present study, the 3D structures of the fuel assembly, including UO2 fuels, claddings, and channel box, and lower tie plate, were modeled. The neutron production in fuels was evaluated for each burnup step, which considered the operational conditions and burnup history, including the cooling time. Since the burnup of the fuel itself was calculated through this process, the neutron flux and neutron spectra varied and were evaluated at each step. The neutron transport from the fuels in the cladding through the lower tie plate was calculated within these 3D structures using the Monte Carlo method for each step. The neutrons lost their energies by moderators, subsequently, activating the material. The nuclear cross-section data were prepared based on the Japanese evaluate nuclear data library, JENDL 4.0 [21].

As the input data for the above activation calculation, cobalt (59Co) impurity, a parent of 60Co, in the irradiated tie plate was directly obtained after the metal specimen was dissolved in HF-HNO3 solution and diluted using the inductively coupled plasma–atomic emission spectrometry (Seiko Instruments Inc., SPS3500DD).

Results and discussion

Table 1 presents the 60Co specific activity results obtained using ɤ-spectrometry at each sampling position shown in Fig. 1. The activities were decay corrected at shutdown (half-life; 5.27 years) and were found to decrease to 1/5th value from the top to bottom of the tie plate away from fuels as a neutron source. These radioactivities in the range of 107–108 Bq/g were comparable to 7.9 × 107 Bq/g of an activated steel nut [9] and 1.2 × 108 Bq/g of a reactor upper internal component [12].

Table 1 Specific activity of 60Co for irradiated lower tie plate obtained using ɤ-spectrometry and the ratios of experimental and calculated (E/C) for two calculation methods. Sampling position is shown in Fig. 1. Decay correction was conducted at the end of irradiation (May 2000)
Full size table

The Co impurity content in the irradiated tie plate was obtained experimentally as an input for accurate activation calculations. Because of commercial confidentiality, the numerical value of Co content is not disclosed and was larger than 176 ppm determined for the activated steel nut [9] by a factor of approximately 2.16.

The radioactivity calculated using ORIGEN2.2-UPJ code [22] was 4.55 × 108 Bq/g, where the neutron flux at the lower tie plate set to 1/4th of the fuel-loaded part as assumed in the safety case [1, 2]. The experimental 60Co activities were lower than the calculated activities as indicated by the experimental and calculated ratios (E/C) in Table 1. Particularly, the E/C ratios for ORIGEN2 in the lower area (G2–G5) were small (< 0.1), suggesting that the assumption of neutron flux in safety case was too conservative.

Unlike the simple ORIGEN calculation, the activation analysis results acquired via MCNP-BURN2 showed good agreement with the experimental data in the E/C range 0.57–0.89. Since the initial composition (Co impurity) and cross section were the same, neutron conditions were considered to affect the difference in both calculations. While a constant neutron fluence and energy spectrum were assumed in ORIGEN, the MCNP-BURN2 calculated their changes by calculating the sequential nuclear fission reaction at the reactor core. Further changes in the neutron flux and spectrum through tie plate were followed and, thus, the subsequent neutron irradiation to the tie plate components were calculated by incorporating all phenomena and reactions. However, a similar MCNP model simulation, where the Co content was determined as 176 ppm, reported an opposite trend of E/C = 2.3 (E = 79,450 kBq/g, C = 34,784 kBq/g) [9]. The different trend of E/C is not easy to understand. One possible reason could be the low Co content (176 ppm), which was approximately half of that determined in this study as mentioned previously. Moreover, the complicated and difficult-to-verify neutron calculations may have had a greater impact, because similar E/C trends of 4.0 and 2.7 for 14C and 54Mn were also reported [9].

Figure 1 shows the radioactivity imaging of 60Co using IP measurements. The color gradation revealed the qualitative 60Co distribution in the tie plate corresponding to the ɤ-measurement results shown in Table 1. Figure 2 presents the relative values of 60Co activity vertically distributed through the tie plate obtained by each method as IP, ɤ-spectrometry, and activation calculation using MCNP-BURN2. The activities for IP measurements at the beginning of A and B were corrected to match those for ɤ-measurements at G1 and G2, respectively. The MCNP-BURN2 simulation activities decreased by a quarter with increasing distance from the top as shown in the ɤ-measurement results in Table 1. Thus, the neutron flux was estimated to decrease from 1/5th to 1/4th value while passing from the top to bottom of the tie plate.

Fig. 2
figure 2

Vertical distribution of 60Co in the lower tie plate comparing the measurements and the MCNP-BURN2 results

Full size image

As mentioned previously, the MCNP-BURN2 results showed good agreement with the experimental data. Therefore, the 3D distribution model was constructed using MCNP-BURN2 to visualize the 60Co activity in the lower tie plate (Fig. 3). The 60Co activities were corrected by the decay constant at the end of irradiation (May 2000). Subsequently, the vertical distribution was well reproduced, whereas the horizontal distribution was not significant, because the neutron source was at the top.

Fig. 3
figure 3

3D model of 60Co distribution in the lower tie plate

Full size image

This developed 3D modeling will be applied to estimate the evolution of radionuclide release behavior by coupling with corrosion rate of stainless steel. This source term modeling for safety case will be considered in future research. Since the corrosion behavior under repository conditions is being presently studied [23, 24], modeling the radionuclides leaching from tie plate could be easily conducted. Further, other important but difficult-to-measure radionuclides, such as 14C and 36Cl, will also be investigated in the future.

Conclusion

As a source term study of activated metallic waste, the content and distribution of 60Co were investigated in the irradiated lower tie plate using the IP method and ɤ-spectrometry. The 60Co activity at the shutdown was 107–108 Bq/g, with a fivefold difference between the top and bottom. Despite using experimentally obtained Co impurity, the conventional activation calculation used in the safety case was conservative but improved using the MCNP-BURN2 code and by calculating the changes in the neutron flux and spectrum in the system. Finally, the 3D distribution for the lower tie plate was modeled for future research to develop radionuclide leaching as the source term model in a safety case. Moreover, difficult-to-measure but important radionuclides, such as 14C and 36Cl, are being investigated.