In the present study, 129I activities and 129I/127I atom ratios were measured in 60 soil samples contaminated by the Fukushima Daiichi Nuclear Power Plant (FDNPP) accident. The 127I concentrations, 129I activities, and 129I/127I atom ratios in dry-weight were observed to be 0.121–23.6 mg kg−1, 0.962–275 mBq kg−1, and (0.215–79.3) × 10−7, respectively. The maximum values of both 129I activities and 129I/127I atom ratios in Japanese soil increased about three orders of magnitude due to this accident. The equation logy = 0.877logx + 0.173 (Pearson’s r = 0.936; x, 129I concentration; y, 131I concentration; decay-corrected to March 11, 2011) instead of a simple constant may be a better way to express the relationship between 129I and 131I in Japanese soil affected by both global fallout and FDNPP accident fallout. In addition, a moderate correlation was observed between 129I and 135Cs (logy = 0.624logx + 1.01, Pearson’s r = 0.627; x, 129I activity; y, 135Cs activity). However, 129I presented larger fractionations with less volatile radionuclides, such as 236U, 239Pu, and 240Pu. These findings indicated 135Cs could be roughly estimated from 129I or 131I; this is advantageous as fewer 135Cs data are available and 135Cs/137Cs is being considered a promising tracer during radiocesium source identification.
The Fukushima Daiichi Nuclear Power Plant (FDNPP) accident in 2011 released massive amounts of radionuclides into the terrestrial environment, including both short-lived radionuclides (e.g. 133Xe, 5.2 d; 131I, 8.0 d; 133I, 20.8 h; 134Cs, 2.1 y) and long-lived radionuclides (e.g. 129I, 1.57 × 107 y; 135Cs, 2.3 × 106 y; 236U, 2.342 × 107 y; 239Pu, 24,110 y and 240Pu, 6,564 y)1,2. The radionuclides with shorter half-lives have higher specific radioactivities, making them suitable for conventional radiometric methods3. Furthermore, conventional radiometric methods, such as γ spectrometry, are easier to apply as they do not have complicated procedures for sample treatment, chemical separation, and purification. Therefore, those short-lived radionuclides with high radiation exposure risk, such as 131I and 134Cs, were extensively studied in the initial stage of the FDNPP accident. On the other hand, the data for long-lived radionuclides, such as 129I, 135Cs, 236U, 239Pu, and 240Pu related to the FDNPP accident are limited4.
Investigation of the distributions and their relevance of 129I, 135Cs, 236U, 239Pu, and 240Pu in Japanese soil is highly required for three major reasons. First, 129I, 135Cs, 236U, and 239+240Pu were indeed released from the FDNPP accident, and their amounts were estimated to be (0.66–5.5) × 1010, 6.74 × 1010, 1.2 × 106, 2.3 × 109 Bq, respectively1,5,6. Second, these long-lived radionuclides can be determined in environmental samples for a long time after a nuclear accident; therefore, they have great potential to act as proxies for short-lived radionuclides that are of greater radiological concern. Initial reconstructions of the distribution of 131I deposition through the measurement of 129I were obtained, as it was known that there were strong correlations between 131I and 129I activities in the contaminated surface soil samples affected by both the Chernobyl accident7 and the FDNPP accident8,9,10,11. Third, they are well-suited tracers with great potential for source identification. 135Cs/137Cs has been proved to be a powerful alternative tracer of 134Cs/137Cs for radiocesium source identification to overcome the drawback of the short half-life of 134Cs in studies of the FDNPP accident5,12,13,14. Significant work has been done that revealed the release of trace amounts of Pu during the FDNPP accident and proved 240Pu/239Pu is a good tracer for Pu source identification15,16,17. At the same time, many scientists have been trying to find evidence of 236U release during the FDNPP accident6,18,19,20,21. Finally, Shinonaga et al.19 and Yang et al.21 were able to present evidence of 236U release from this accident in aerosol samples and soil samples, respectively.
Different elements have distinct physicochemical properties, which will result in different dispersal and deposition processes from reactor cores to the environment at the initial stage of their emissions and different post-depositional redistributions (vertical diffusion or migration) in soil samples. For example, evidence was observed that 129I migrated downward more rapidly in soil than 137Cs did after both the Chernobyl accident22 and the FDNPP accident23. Therefore, it is better to apply a long-lived isotope to act as a proxy for a short-lived isotope of the same element, such as 129I for 131I, and 135Cs for 134Cs and 137Cs. However, it should be noted that only trace amounts of 129I, 135Cs, 236U, 239Pu and 240Pu were released from the FDNPP accident, and background activities before this accident account for the major fraction of these radionuclides, as revealed by previous studies and mentioned above. For example, before the FDNPP accident, 129I was already present in Japanese soil owing to natural generation (cosmic ray reactions with Xe, spontaneous fission of 238U, thermal neutron induced fission of 235U and Te) and human nuclear activities (atmospheric nuclear test since 1945, discharge from spent-nuclear-fuel reprocessing plants, fallout from the Chernobyl accident). Multiple sources and long-term redistributions make their relationships complicated. However, most reports only focused on one or two long-lived radionuclides in each study. Therefore, it is necessary to build a database to show data for as many as possible long-lived radionuclides in the same samples to establish their signatures and illustrate the differences and relevance among them in soil samples affected by the FDNPP accident fallout and global fallout.
Unfortunately, there are still not sufficient data related to the FDNPP accident for environmental samples from Fukushima Prefecture to provide regional information on the deposition of multiple long-lived radionuclides, such as, 129I, 135Cs, 236U, 239Pu and 240Pu. The major reason for the limited numbers of data of long-lived radionuclides is the challenge in their measurement. Since mass spectrometric methods are more sensitive for longer-lived radionuclides3, inductively coupled plasma - mass spectrometry (ICP-MS) is the most widely applied method to measure Pu isotopes15,16,17. For 129I and 236U in environmental samples, accelerator mass spectrometry (AMS) is the most widely applied and it offers the highest sensitivities6,18,19,20,21,24,25,26. However, due to the high instrument cost, there are only about 110 AMS facilities worldwide, and most of them are mainly applied to the routine analysis of 14C for dating purposes; only about ten can be used to study 236U20,21,25 and only about 22 AMS facilities can be used to study 129I25. Therefore, we have developed novel methods for rapid measurement of 135Cs, 129I and 236U in environmental samples with high throughput that are compatible with the advanced triple-quadrupole ICP-MS (ICP-QQQ)14,20. Compared with AMS, ICP-QQQ instruments are relatively inexpensive and can be afforded by ordinary laboratories, giving ICP-QQQ bright prospects in the field of long-lived 135Cs, 129I and 236U applications in the future.
Here, we report data for 127I, 129I, and 131I in 60 soil samples, with heavy 134Cs contamination due to the FDNPP accident, that were collected immediately after this accident. For those soil samples without 131I activity data, 131I activities were reconstructed via deduced 129I-131I equation and measured 129I activities. In addition, other radionuclides (major long-lived ones), such as 135Cs, 236U, 239Pu, 240Pu, were also considered with regard to the differences and relevance among them in soil samples affected by the FDNPP accident fallout and global fallout, in order to see whether 129I can track other radionuclides (135Cs, 236U, 239Pu, and 240Pu) derived from the FDNPP accident fallout and global fallout.
These 60 soil samples were collected in Fukushima Prefecture and were heavily contaminated with 134Cs (12.9–1.10 × 105 Bq kg−1). Since 134Cs (t1/2 = 2.06 y) in the environment before the FDNPP accident has decayed out to an undetectable level5,14,27, these high values indicated significant radiocesium contamination due to the FDNPP accident. As shown in Supplementary Table S1, the 127I concentrations, 129I activities, and 129I/127I atom ratios in dry-weight were observed to be 0.121–23.6 mg kg−1, 0.962–275 mBq kg−1, and (0.215–79.3) × 10−7, respectively. In addition, the activities of 135Cs, 236U, 239Pu, and 240Pu in soil were in the ranges of 4.55–376, 0.005–0.244, 4.26–227, and 2.76–144 mBq kg−1, respectively.
Presently, no reliable value is available for 129I/127I atom ratio in the terrestrial environment prior to the start of the nuclear era. Before the FDNPP accident, surface soil and atmospheric fallout collected far from the Tokai Reprocessing Plant (TRP) in Japan had 127I concentrations in somewhat over lapping ranges of 0.4–55.328,29,30,31,32 and 2.79–11033,34 mg kg−1, respectively; the 129I activities were in the ranges of <0.06–4.5528,29,30,31,32 and 0.157–65.733,34 mBq kg−1, respectively; and the 129I/127I atom ratios were relatively low, being in the ranges of (<0.1–2.18) × 10−8 28,29,30,31,32 and (0.008–2.43) × 10−7 33,34, respectively. For two surface soil samples collected in 2008 and 2009 within 10 km to the west of the FDNPP, their 129I activities and 129I/127I atom ratios were observed to be low, being 0.339–0.422 mBq kg−1 and (1.40–1.57) × 10−8, respectively29,30. However, after the FDNPP accident, the 129I activities and 129I/127I atom ratios in soil increased sharply and presented values of 0.109–160 mBq kg−1 and 3.39 × 10−9–1.01 × 10−5, respectively8,9,10,11,29,30,35,36. In other words, as shown in Fig. 1, after the FDNPP accident, the maximum values of both 129I activities and 129I/127I atom ratios in Japanese soil samples increased about three orders of magnitude relative to the results of the present study and previous studies as mentioned above. These indicated that the amount of radioiodine released in the FDNPP accident was significant. Owing to the short half-life of 131I, it has a high specific radioactivity. Therefore, apart from 129I, it is also vital to obtain 131I activity values to study the radiation exposure risk from 131I to the local population and the environment in this accident.
Due to the short half-life of 131I, we could only determine its activities (decay-corrected to collection dates, dry weight) in 22 soil samples, and they were in the range of 1.36–81.3 kBq kg−1. Since 129I and 131I have the same chemical and environmental behaviours, and similar production routes in a nuclear reactor, 129I could be an ideal proxy for 131I. In soil samples contaminated by the Chernobyl accident, a strong linear correlation was found between 131I and 129I activities with a 129I/131I atom ratio of 15.2 ± 4.7 (decay-corrected to April 26, 1986)7. Simple slope constants of linearity regressions were also used to reconstruct the 131I level and distribution pattern by long-lived 129I during the FDNPP accident8,9,10,11. However, to date, limited 129I/131I data have been obtained and large errors (most >20%) were observed in previous studies. Four reasons may explain this phenomenon: 1) decreased 131I activities by the time of measurement, 2) the time-consuming sample treatment needed for 129I analysis by AMS, 3) the limited numbers of AMS systems available, and 4) the light contamination of most regions by the FDNPP accident fallout, but with a large background contribution from multiple sources8,9,10,11. For example, Miyake et al.10 reported a relative deviation of 22.1% for 129I/131I atom ratios in 50 soil samples; Muramatsu et al.11 even reported a relative deviation of 48.4% for 129I/131I atom ratios in 82 soil samples. In addition, a higher relative deviation (49.7%) was also observed in the present study. It is vital to compile as many 129I/131I data as possible to obtain a comprehensive relationship between 129I and 131I in soil affected by the FDNPP accident fallout and global fallout. However, if more samples with strong background (global fallout) contribution are added, the relative deviation of 129I/131I atom ratios will increase sharply as mentioned above. Therefore, only using simple slope constants of linearity regressions is not an ideal way to express the relationship between 129I and 131I, since the region heavily contaminated by the FDNPP accident is restricted to a narrow strip.
We used the available 129I and 131I data in soil samples, affected by the FDNPP accident fallout and global fallout, in previous studies together with the data in the present study to illustrate a more comprehensive relationship between 129I and 131I. After a linear regression analysis, the slope of the line indicates the 129I/131I atom ratio is 26.1 ± 15.5 (decay-corrected to March 11, 2011) and there is no drop in the uncertainties. This indicates again that a simple slope constant of linearity regression cannot explain the relationship between 129I and 131I comprehensively. After linear regression of all data in Fig. 2 expressed as common logarithms, the equation of logy = 0.877logx + 0.173 (Pearson’s r = 0.936; x, 129I concentration (atoms kg−1); y, 131I concentration (atoms kg−1); decay-corrected to March 11, 2011) could be obtained corresponding to the linear regression line. The standard errors for slope and intercept of this equation were low, being 0.030 and 0.33, respectively. The slope <1 in the log-log plot clearly points out systematically varying isotope ratios. Fujiwara et al.8 reported that the contributions of background 129I in five topsoil samples collected in Tsukuba, Japan, about 170 km southwest of the FDNPP, ranged from 38.9% to 41.4%. However, Matsunaka et al.29,30 observed smaller contributions of background 129I in two topsoil samples collected within 10 km to the west of the FDNPP (<5%) due to the heavy contamination from the FDNPP accident. All these indicated the complicated relationships of the sources and contributions for iodine isotopes in the studied soil samples. In addition, Nishihara et al.37 carried out a model calculation for the radionuclides in the three FDNPP reactors using the ORIGEN2 code, and they estimated significantly distinct 129I/131I atom ratios at the time of this accident which were 31.4, 21.9, and 20.8 for reactor Units 1, 2, and 3, respectively. Therefore, a larger error in more samples for the 129I/131I atom ratio may indicate the significant influence by background 129I in some soil samples and a complicated situation of radioiodine release from the three reactor units during the FDNPP accident. Anyway, presently, using 129I as a proxy for 131I offers the best way to estimate the radiation exposure risk from 131I to the local population and the environment. Therefore, at present, the equation of logy = 0.877logx + 0.173 (Pearson’s r = 0.936; x, 129I concentration (atoms kg−1); y, 131I concentration (atoms kg−1); decay-corrected to March 11, 2011) instead of a simple slope constant of linearity regression with large uncertainty may be a better way to express the relationship between 129I and 131I in the soil samples in Fukushima Prefecture affected by both global fallout and the FDNPP accident fallout.
As shown in Fig. 3 (decay-corrected to March 11, 2011, to facilitate comparison with other studies), the reconstructed 131I activities were 0.023–60.3 kBq kg−1 (decay-corrected to collection dates, dry weight); and, higher 131I activities were observed in the northwest direction from the FDNPP, in agreement with the observation that the radionuclides were predominately deposited northwest of the facility in a band approximately 40 km in length11. It should be noted that only a few samples were collected in the southwest direction from the FDNPP; therefore, in the future more samples from this area are required to show the distribution of 131I and correlation between 129I and 131I more exactly.
After the FDNPP accident, more than 99% of the released activity was due to radionuclides of the elements Kr, Te, I, Xe, and Cs; however, little work had been done on monitoring of radionuclides other than the short-lived 131I, 132Te, 134Cs, 136Cs, and 137Cs1. Radionuclides such as those of less volatile elements (e.g., Pu) and radionuclides with very long half-lives (e.g., 135Cs, 129I, and some actinides such as 236U) have been understudied by comparison4,5,13,16,21. Therefore, in the present study, we combined the data for 135Cs, 236U, 239Pu, and 240Pu in the same samples with the data of the radioiodine isotopes, trying to see whether 129I or 131I could trace other radionuclides in soil samples. To the best of our knowledge, this is the first time these long-lived radioisotopes have been simultaneously compared in the same samples. As shown in Fig. 4, 135Cs and 129I activities had a moderate linear correlation (logy = 0.624logx + 1.01, Pearson’s r = 0.627; x, 129I activity (mBq kg−1); y, 135Cs activity (mBq kg−1)). Very recent studies have proved that 135Cs/137Cs is a potential tracer of radiocesium in environmental science, however, the data of 135Cs in soil samples are few in number and no 135Cs data in the Japanese environmental samples before the FDNPP accident are available5. Although in the course of the FDNPP accident, Matsunaka et al.30 found that the 129I/137Cs ratios were varying by several orders of magnitude, we consider it feasible to roughly estimate 135Cs activities in soil samples using published 129I activities, under the special conditions that significantly fewer numbers of 135Cs data in soil are available compared with 129I data at present and most fractions of 135Cs in the soil are from global fallout rather than the FDNPP accident fallout5. In addition, the 236U/129I, 239Pu/129I, and 240Pu/129I activity ratios varied largely from 7.84 × 10−5 to 1.23 × 10−1, from 0.144 to 72.9, and from 0.102 to 49.7, respectively. 129I did not show significant correlations with 236U, 239Pu, and 240Pu. The major reason may be that 1) the releases of 236U, 239Pu, and 240Pu from the FDNPP accident were trace amounts compared with the previous depositions of the radioiodine isotopes; 2) most of the 236U, 239Pu, and 240Pu present in Japanese soil samples are from global fallout4,5,13,16,21; 3) 129I and 135Cs are volatile radionuclides, while 236U, 239Pu, and 240Pu are less volatile radionuclides. Therefore, the differences between 129I and 236U, 239Pu, or 240Pu become larger than that with 135Cs after release. From the present study, we have confirmed that 129I in Japanese soil could roughly track 135Cs, rather than 236U, 239Pu, and 240Pu derived from the FDNPP accident fallout and global fallout.
It should be noted that the uncertainty of ICP-QQQ is somewhat larger than that of AMS at present. However, the amount of 129I in the unprocessed spent nuclear reactor fuel is 10 times more than that released to the environment and more 129I will be produced with the increasing number of nuclear power reactors being built. If most of the spent fuel is going to be reprocessed and 129I is released to the environment, it may increase the ratio of 129I/127I to 10−3 38. Then, the ICP-QQQ method will become a more feasible method to investigate 129I in the environment with smaller uncertainties.
The procedure details for soil sampling and pre-treatment have been described elsewhere27. Surface soils (0–5 cm) were collected from 60 sites in Fukushima Prefecture (Fig. 3 and Supplementary Table S1) during four sampling expeditions conducted in 2011, from April 12 to 16, April 26 to 28, June 6 to 10, and June 15 to 16. The collection sites were mainly restricted to the heavily contaminated region where the radioactive plume due to the FDNPP accident was washed out by rainfall. Fukushima Prefecture is divided by mountain ranges into three regions (from west to east) showing large temperature and weather contrasts21. On average, annually, central Fukushima receives 1166 mm of precipitation and 189 cm of snow, respectively.
After large particles and plant roots were removed by handpicking, soil samples were transferred into 100-mL polystyrene containers, and then, only the fine fraction of soil particles (diameter below 2 mm) was analyzed.
Reagents and Materials
Ultrapure grade 25% TMAH (TAMAPURE-AA) was obtained from Tama Chemicals (Kawasaki, Japan). Analytical grade (NH4)2SO3 solution (0.6–1%), 0.1 mol L−1 KI solution, CCl4, and NaNO2 were obtained from Kanto Chemical Co. Inc. (Tokyo, Japan). Single-element standard solutions (1000 mg L−1) of Cs, Mo, Cd, In, Li, Mg, Y, Ce, Tl, and Co were also purchased from Kanto Chemical Co. Inc. K2S2O8 was obtained from Wako Pure Chemical Industries, Ltd. (Osaka, Japan). High purity O2 (>99.999%) and Ar (>99.999%) were supplied by Taiyo Nippon Sanso Corp. (Tokyo, Japan). IAEA reference material (soil, IAEA-375), NIST standard reference materials (marine sediment, NIST SRM 4357), and GSJ Geochemical reference materials (rock, JB-2 and JB-3; stream sediment, JSd-3) were used for iodine method validation, as shown in Supplementary Tables S2 and S3.
Analysis of Radionuclides
An Agilent 8800 (ICP-QQQ, Agilent Technologies, Santa Clara, CA, USA), featuring an octopole collision/reaction cell situated between two quadrupole mass filters (first, Q1; second, Q2), was employed for 129I/127I atom ratio analysis. Pure O2 (>99.999%) was introduced into the collision/reaction cell via the No. 4 cell gas line. A high efficiency sample introduction system (APEX-Q, Elemental Scientific, Omaha, NE, USA) equipped with a PFA MicroFlow nebulizer was used as the sample introduction system. The ICP-QQQ was optimized on a daily basis using 1 ng mL−1 of standard solution containing Ce, Co, Li, Mg, Tl, and Y in 0.01% (NH4)2SO3. The optimized operation parameters are summarized in Supplementary Table S4. 133Cs was measured as internal standard by ICP-QQQ single MS mode for 127I concentration measurement. In addition, 95Mo, 111Cd, and 115In were monitored to check the contributions of their polyatomic interferences at m/z 129 during the analysis of 129I by ICP-QQQ. Activity of 131I was determined by γ-ray spectroscopy (ORTEC GEM-40190, Seiko-EG&G, Tokyo, Japan) at an energy of 636 keV from 1000 to 80,000 s27. Mixed gamma standard sources with different sample heights were used for efficiency correction; they were obtained from the Japan Radioisotope Association. Because the measurements were started in the middle of May 2011, 131I could only be determined in 22 soil samples.
TMAH Exaction and Purification of Iodine for ICP-QQQ Analysis
The procedures for iodine extraction from solid environmental samples and purification (solvent extraction and back-extraction) for ICP-QQQ analysis are summarized in Supplementary Fig. S1. Six steps are shown here briefly. (1) 10% TMAH extraction: eighteen of soil samples (1 g) were weighted into 50 mL centrifuge tubes, and then incubated for iodine extraction by 10% TMAH solution (25 mL). Incubation was done in an aluminium block bath (DryThermoUnit DTU–2CN, TAITEC Corp., Koshigaya, Japan) at 90 oC for 2 h. (2) Iodine release from organic matter: after centrifugation at 1000 rpm for 5 min (these same parameters for centrifugation were used in the following sections), the aqueous phase was transferred to a new centrifuge tube. A 0.1 mL aliquot of the aqueous phase was taken out, and diluted with 0.01% (NH4)2SO3 to ensure the measured 127I concentrations remain within the range of the calibration curve (0.01–5 ng mL−1). At the same time, Cs was added to get a final concentration of 5 ng mL−1 as internal standard for ICP-QQQ single MS mode analysis since the extracted Cs amount was ignorable compared with the added amount. For the remaining extracted TMAH solution, after adding K2S2O8 (about 0.03 g), the solution was incubated at 60 °C in the aluminium block bath overnight to convert organic iodine to inorganic iodine (iodate, IO3 −). Then, the precipitate was discarded after centrifugation. (3) IO 3 − reduction: CCl4 (5 mL) was added to avoid element iodine release during the following chemical reaction. After adding 1% (NH4)2SO3 (1.5 mL) and then adding 6 M HNO3 (9 mL) to adjust pH < 3, IO3 − was reduced to iodide (I−) by (NH4)2SO3 (IO3 − + 3SO3 2− → I− + 3SO4 2−). (4) Organic layer removal: after mechanical shaking for 2 min and then centrifuging, the organic matter became a thin layer between the TMAH phase and the CCl4 phase. The upper TMAH phase and the bottom CCl4 phase were transferred to another 50 mL centrifuge tube, and the black organic matter was discarded. (5) I 2 transformation: I− was oxidized to elemental iodine (I2) by addition of 5% NaNO2 (0.4 mL) under acidic conditions (NO2 − + 2I− + 4 H+ → I2 + 2NO + 2H2O). At the same time, I2 was extracted with CCl4 (5 mL) by mechanical shaking for 2 min. After centrifugation, the organic phase was separated from the aqueous phase for the next procedure. Elemental iodine was extracted again with CCl4 (3 mL). (6) I − transformation and extraction: I2 in the organic phase was reduced and extracted back into 0.01% (NH4)2SO3 solution (1.5 mL) as I− (I2 + H2O + SO3 2− → 2 H+ + 2I− + SO4 2−) (back-extraction). Since the shaking during back-extraction was not done strongly, no emulsion appeared in the bottom phase; this helped to increase iodine recovery. Finally, the 129I/127I atom ratio was analyzed by ICP-QQQ MS/MS mode. For 1 g soil samples, the method detection limits for 127I and 129I in ICP-QQQ MS-MS mode were 3.80 ng kg−1 and 2.62 × 10–4 Bq kg−1, respectively. This method could measure 129I/127I > 10−8 accurately, as shown in Supplementary Tables S2 and S3.
In addition, the data of 135Cs, 236U, 239Pu and 240Pu obtained in our previous studies5,14,20,21, were combined with the present data of 129I and 131I as a means to illustrate the differences and relevance among them in soil samples affected by the FDNPP accident fallout and global fallout, in order to see whether 129I can track other radionuclides (135Cs, 236U, 239Pu, and 240Pu) derived from the FDNPP accident fallout and global fallout. Details about the analysis of 135Cs, 236U, 239Pu and 240Pu can be found in the Supplementary Information.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We wish to express our gratitude to Drs Tokonami, Hosoda, Sorimachi, Nakata, and Kasai for their cooperation in the sampling. This work was supported by JSPS KAKENHI (Grant Numbers: 24110004, 24310002, 16K12592). G.Y. thanks the National Natural Science Foundation of China for financial support to carry out experiments abroad (Grant Numbers: 21407149, 11435002). G.Y. thanks Hirosaki University for a grant for Exploratory Research by Young Scientists and Newly-appointed Scientists.