Production scheme for diagnostic-therapeutic radioisotopes by accelerator neutrons

Interest has been growing in the development of medical radioisotopes used for noninvasive nuclear medicine imaging of disease and cancer therapy. Especially the development of an alternative production scheme of 99Mo, the mother radioisotope of 99mTc used for imaging, is required, because the current supply chain of the reactor product 99Mo is fragile worldwide. We have proposed a new production scheme of 99Mo as well as therapeutic radioisotopes, such as 64Cu and 67Cu, using accelerator neutrons provided by the natC(d,n) reaction. Based on this scheme we have obtained high-quality 99mTc, 64Cu, and 67Cu suitable for clinical use by developing both production and separation methods of the radioisotopes. We proposed a new facility to constantly and reliably produce a wide variety of high-quality, carrier-free radioisotopes, including 99Mo, with accelerator neutrons. We report on the development of the proposed scheme and future prospects of the facility toward the domestic production of medical radioisotopes.


Introduction
More than 150 different radioisotopes (RIs) have been used in medical, industrial, research as well as commercial applications, such as food processing, agriculture, and structural safety. 1) In these applications, the radioindicator tracer method, which was discovered by George de Hevesy, has been playing an important role.
1.1. Discovery of a radioindicator tracer concept. George de Hevesy, known as the founder of radioanalytical chemistry and nuclear medicine, published a paper in 1913 claiming that radio-indicator tracers play a unique role in chemical analysis. 2) The concept of radioindicator tracer chemistry was born through a talk with Ernest Rutherford. 3) He encouraged Hevesy to separate the natural radioisotope of lead-210 ( 210 Pb) from its admixture with large amounts of nonradioactive lead stored at that time in Rutherford's laboratory. Hevesy used chemical methods for such separation. After failing at this, Hevesy had an idea to use 210 Pb as an indicator of lead. Hevesy studied the transfer of lead from soil in different parts of bean plants using 210 Pb, which was the first application of the radioactive tracer technique to biology. 3) The extreme sensitivity of physical radioassay methods allowed him to carry out these experiments with such minuscule concentrations of lead so as to avoid its toxic properties. In 1935 Hevesy investigated the distribution and kinetics of the exchange of phosphorus in different parts of animals using the 32 P radioisotope, which was the first radioindicator study in life sciences. 3) The essence of the radioindicator tracer method is that radioactivity is of such minute quantities that it will not cause any toxic affect on the system, and can be detected with high sensitivity by discriminat-ing any conceivable background events as low as possible. A radiolabeled tracer allows noninvasive measurements of the distribution and function in a biological system. Thus, a door to so far unexplored nuclear medicine had been opened.
1.2. Medical radioisotopes for diagnostics and therapy in nuclear medicine. Medical RIs in the form of radiopharmaceuticals have been used for noninvasive diagnostic imaging studies of diseases and therapeutic applications in cancer treatments, resulting to account for the majority of the applications of RIs. 1) Recently, an approach involving the merging of therapeutic and diagnosis (imaging) treatments, called the "theranostics" approach, has become very important to make a personalized medicine treatment for a specific patient. 4) Personalized medicine is a form of medicine that uses information about a patient's own genes or proteins to prevent, diagnose, or treat some disease. A personalized medicine treatment in nuclear medicine is being made using theranostic radioisotopes in the same patients as follows. Firstly low-dose molecular imaging is performed to obtain necessary pre-treatment information concerning the patient's biodistribution and dosimetry in patients, and secondly higher-dose target molecules therapy is carried out. 5) In order to promote this theranostics approach while aiming at personalized medicine treatment, the development of a variety of medical radioisotopes for the imaging and therapy of various diseases (especially cancers) is essential. The term "theranostics" was coined by Funkhouser. 6) Radioisotope imaging is divided into two modalities based on the type of decay and resultant particle emission and detection. 4) Single-photon emissioncomputed tomography (SPECT) is performed using .-ray-emitting radioisotopes with energy below about 400 keV, which can penetrate the patient's body and be detected by a gamma camera. Technetium-99m ( 99m Tc) with a half-life (T 1/2 F 6 h) is the most widely used radioisotope in diagnostic imaging studies with SPECT. 4) Positron emission tomography (PET) imaging is the second-most common imaging modality with the use of a positron (O D )-emitting radioisotope. Fluorine-18 ( 18 F) with T 1/2 F 1.8 h has excellent nuclear properties for a PET radioisotope to provide optimal resolution, and an appropriate half-life for making radiopharmaceuticals. 4) Cancer treatments have been achieved by performing surgery, chemotherapy, radiation therapy, immunotherapy, targeted therapy, and hormone therapy etc. Radiation therapy is carried out using a certain amount of radiation doses to kill cancer cells and shrink tumors. It is divided into external radiation therapy, which is performed using such particles as X-rays, electrons, protons, and carbons, and internal radiation therapy which is carried out using therapeutic radiopharmaceuticals, including radioisotopes and pharmaceuticals, by employing techniques involving either brachytherapy or radioimmunotherapy. 5) In brachytherapy, radioisotopes are placed inside or near tumors of patients to treat cancer cells. Note that growing research activities include combining radiopharmaceuticals with conventional treatments, such as chemotherapy and external beam radiotherapy, that would need a variety of medical RIs to form effective radiopharmaceuticals.
Medical RIs used for internal radiation therapy should have the following physical and chemical properties 4) : firstly, physical properties, such as the physical half-life (T 1/2 ), particle decay mode (O ' -ray, Auger electron, .-ray, ,-particle emission), and the energy of the emitted particle are important. The half-life of RIs must match with the pharmacokinetics of the radioactive drug for uptake and clearance from normal versus targeted (disease site) tissues. It is a crucial point for any radiopharmaceuticals to maximize the dose to the target tissue and minimize the dose to normal tissue. An excellent medical radioisotope used for diagnostics and therapeutics has a short half-life, typically within several hours, like 18 F, and within ten days, respectively. Secondly, the chemical properties should be suitable for allowing the radioisotope to be incorporated into all sorts of molecules, including such as protein compounds for labeling the RIs with pharmaceuticals having high radiochemical yields and a high radionuclide purity. Thirdly, we have to prepare carrierfree RIs (without any other isotope of a sample nuclide) with high specific activity, because a typical activity of a 99m Tc radiopharmaceuticals solution administered to a patient is as high as about 740 MBq/(a few ml). 4) Note that the specific activity, which is defined as the activity per quantity of the atoms of a particular radionuclide, is usually given in units of Bq/g. expected to be high because carrier-free RIs can be obtained by a chemical separation process owing to the fact that an atomic number of the RIs differs from that of 235 U. The specific activity of the RIs produced by the latter reaction is mostly very low because an atomic number of the RIs is the same as that of a sample, and therefore the RIs are not possible to be chemically separated from the sample. Neutron-rich nuclei decay by emitting O ! -rays and .-rays, and therefore radiopharmaceuticals containing O ! -ray emitting RIs are used for radioimmunotherapy (RIT), and those containing .-ray emitting RIs are used to diagnose the dynamics in a living body via SPECT. In accelerators, a variety of carrier-free RIs including many proton-rich light RIs, such as 18 F and 11 C, are produced using mostly proton beams. 5) These RIs decay mostly by emitting O D -rays or alpha particle, which are used in imaging and RIT, respectively. Typical medical RIs produced in reactors and accelerators are listed in Table 1. 7) 1.4. Challenges in developing medical radioisotopes. Currently there are important challenges in the productions of medical RIs used for diagnosis and therapy, respectively. Firstly, the supply chain of 99 Mo (T 1/2 F 66 h) is vulnerable and unreliable, which has frequently caused a shortage of 99 Mo worldwide since 2008. 8) More than 80% of all diagnostic procedures in the world are carried out using 99m Tc obtained from a 99 Mo/ 99m Tc generator. In Japan approximately 0.9 million procedures/year (about 2,750 procedures/day) are performed using 99m Tc. 9) Japan imports all of its required 99 Mo several times per week. Therefore, a reliable and constant supply of 99 Mo is a key issue for ensuring the routine application of 99m Tc worldwide. About 95% of 99 Mo is produced by the fission reaction of enriched 235 U in several aging research reactors around the world. The vulnerable situation of the supply chain of 99 Mo is the impetus to study alternative methods for producing 99 Mo and/or 99m Tc in reactors or accelerators worldwide. 8) In fact, strong efforts have been undertaken to develop an alternative production route of 99 Mo. Until now, as far as we know, no alternative method has yet succeeded to secure a constant supply of 99 Mo. Secondly, in order to promote a theranostics approach aimed at a personalized medicine treatment for a specific patient, the development of theranostic radioisotopes is highly required. Recent great success in the development of therapeutic RI of 177 Lu (in reactors) 10) and 225 Ac (decay product of 229 Th of nuclear waste in reactors) 11) has been prompting further development of a wide variety of medical RIs by innovative production methods.
This review focuses on our studies, started in 2010, that aim to develop a new scheme of medical radioisotopes production and separation to secure a reliable supply chain of 99 Mo for domestic use and to proceed the theranostics approach while aiming at personalized nuclear medicine. We discuss 99m Tc imaging and the supply chain of 99 Mo in chapter 2. The new production scheme of medical radioisotopes, such as 99 Mo,64 Cu and 67 Cu, using accelerator neutrons is introduced in chapter 3. The development of separation methods of 99m Tc from 99 Mo, and 64 Cu and 67 Cu from Zn samples are given in chapter 4. The deuteron accelerator and the accelerator neutron source are described in chapter 5. Conclusions and future prospects are given in chapter 6. The excited state of 99m Tc at 143 keV is populated by the decay of 99 Mo, as shown in Fig. 1. 12) 99m Tc has physical and chemical properties suitable to perform a diagnostic procedure. 4) First, a short half-life of 99m Tc (T 1/2 F 6 h) allows one to use a large quantity of 99m Tc activity with a low radiation dose to a patient for obtaining clear imaging. Second, 141 keV .-rays emitted from 99m Tc to the ground state of 99 Tc are detected by a gamma camera with a high detection efficiency with a low-energy collimator. Third, 99m Tc has a versatile chemistry that allows it to be incorporated into all sorts of molecules. Fourth, 99m Tc is routinely produced in 99 16) When they developed a system to separate 132 I (T 1/2 F 77 h) from its parent 132 Te (T 1/2 F 2.3 h), aiming at medical studies, they happened to detect a trace contaminant, which was later proved to be 99m Tc. It was then realized that 99m Tc was coming from its parent, 99 Mo. Similarities between the chemistry of the tellurium-iodine parent-daughter pair and the molybdenum-technetium pair led to the development of a 99 Mo/ 99m Tc generator system.
Note that Tc, presumably 99 Tc with a relatively short half-life of 2.1 # 10 5 y, compared to the age of the universe of 13.7 billion years, was discovered by P. W. Merrill at the surface of red giant stars in 1952. 17) The observation of Tc provided the first powerful evidence that Tc has been synthesized recently within stars. This was a great contribution for understanding the origins of elements heavier than iron observed in stars. 99 Mo and supply chain of 99 Mo. About 95% of 99 Mo is produced by the fission reaction of enriched 235 U [hereafter: fission- 99 Mo] in a limited number of aging research reactors. In around 2008, just before the shortage of 99 Mo began, five aging nuclear research reactors in The Netherlands, Australia, South Africa, Belgium, and France contributed to meet nearly all (about 95%) of the world's supply of 99 Mo/ 99m Tc. 8) Between 2008 and 2010 a supply crisis of 99 Mo, caused by repeated shutdowns, resulted in many diagnostic tests being cancelled or delayed. 8) The incidents of the reactors highlighted shortcomings and unreliability in the supply of 99m Tc. Currently, 99 Mo has been mostly produced in seven research reactors, listed in Table 2. 18), 19) In addition to concerns related to aging reactors, it is worth noting that the global supply chain of 99 Mo for the 99 Mo/ 99m Tc generator production is complex, as shown in Fig. 2. 20) Namely, after being produced in reactors, 99 Mo is transferred to a processing facility (as listed in Table 3) to be chemically separated and purified. The finished 99 Mo product material is then isolated and shipped to one of eight generator-manufacturing facilities, located in different countries, that supply 99 Mo in the form of a 99 Mo/ 99m Tc generator to end users, such as nuclear pharmacies and hospitals. In addition, the time frame to deliver purified 99  customs, government regulations, flight schedules, weather delays, and pilot refusal. Natural disasters also have the potential to result in significant product shipment delays. A volcano that erupted in Iceland also reminded us of the risk of relying on a small number of 99 Mo production facilities in the world. There are currently only five 99 Mo-processing facilities in the world, as listed in Table 3, in which three of them will end their operations within a decade. 18 99 Mo and 99m Tc supply in both the short (between 2010 and 2017), medium (2017-2025) and long terms (after 2025). Among the actions requested by the HLG-MR was a review of other potential methods for producing 99 Mo in reactors and accelerators.

Production of
The OECD efforts have succeeded to significantly improve the global demand for 99 Mo so as to meet at a near-to-full-service level. However, the fragility of the current production chain of 99 Mo has been remaining, as mentioned above. 18) This vulnerable situation is an impetus to study alternative methods for producing 99 Mo and/or 99m Tc in reactors or in accelerators worldwide. In fact, many efforts are being made for the production of 99 Mo or 99m Tc worldwide. Typical proposed reactions are the (n,.) reaction on 98 Mo or natural Mo samples in reactors and the 100 Mo(p,pn) 99 Mo, 100 Mo(d,p2n) 99 Mo, Table 2. Current irradiation facility for producing 99 Mo around the world. A '6-day curie' is defined as the amount of 99 Mo activity left six days after the generator has left the producer's facilities. 19 99 Mo and 100 Mo(.,n) 99 Mo reactions in accelerators. 22) Reaction routes reported in the OECD report are shown in Fig. 3. 23) 3. New production routes of medical radioisotopes using accelerator neutrons 3.1. 99 Mo production via the 100 Mo(n,2n)- 99 Mo reaction. The current route of fission-99 Mo provides a large amount of 99 Mo with a high specific activity of 9370 TBq/(g 99 Mo) in a single research reactor. Hence, a high specific activity of 99m Tc used for formulating 99m Tc radiopharmaceuticals is obtained from 99 Mo using a commercially available 99 Mo/ 99m Tc generator. Such highly specific activity of fission-99 Mo is provided by high neutron flux of reactors, a large target volume of enriched 235 U, and the high probability of the nuclear fission reaction of 235 U. However, a specific activity of 99 Mo based on any alternative routes, other than that of fission- 99 Mo, is as low as about 1/5,000 of the fission-99 Mo. In fact, all alternative 99 Mo production routes would face the challenges of lower reaction rates and lower specific activity, which must be overcome by fundamental technical breakthroughs. So far, a variety of nuclear reactions have been proposed to produce 99 Mo in accelerators, as discussed in chapter 2. However, a production scheme of medical radioisotopes with fast neutrons from an accelerator (hereafter accelerator neutrons) without using a fissionable element U sample had not yet been considered.
In 2009, we proposed a new route to produce 99 Mo by the 100 Mo(n,2n) 99 Mo reaction using accelerator neutrons provided from an accelerator. 24) In proposing a scheme of 99 Mo production using this route, we considered the following points. Firstly, any scheme of 99 Mo production is required to meet all or a significant part of the domestic demand of 99 Mo from an economic view point. We have kept in mind that because the requirement might be hardly met using existing accelerators, one has to propose the production scheme by taking into account the cost for an infrastructure, including an accelerator. Secondly, the safety and efficacy of the 99m Tc radiopharmaceutical preparation based on the proposed scheme should be ensured. Thirdly, we also took into account a criterion concerning the potential for other isotopes co-production at the same time, which was introduced in the OECD report. 23) It is considered to provide an indication of the economic sustainability, demand risk mitigation and the ability to avoid creating some other isotope shortage. 100 Mo(n,2n) 99 Mo reaction product. In order to meet these requirements, the activity of 99 Mo produced by a single accelerator should be as high as possible by considering the key parameters, such as the nuclear reaction cross section, beam flux, energy, irradiation time, number of sample nuclei, and half-life of the radioisotope, as discussed next. Firstly, the cross section of the 100 Mo(n,2n) 99 Mo reaction has been measured 25) and evaluated 26) by many groups at a neutron energy of around 8 MeV up to 20.5 MeV. The 100 Mo(n,2n) 99 Mo reaction cross section is the largest one (except for an elastic scattering cross section) in the neutron-induced reaction of 100 Mo, about 1.5 barn in the neutron energy (E n ) between 12 and 20 MeV, as shown in Fig. 4. 26) The cross sections of the 100 Mo(n,,), 100 Mo(n,3n), and 100 Mo(n,p) reactions for producing impurity radioisotopes are less than a few mb at E n 9 14 MeV. Hence, 99 Mo could be produced with a minimum level of radioactive waste. Secondly, high-flux accelerator neutrons are expected to be provided by a neutron source based on the deuteron breakup reaction of light elements, such as carbon and beryllium etc. In fact, owing to the progress in accelerator technology, neutrons with a high flux of 10 15 n/s having a most probable energy of 14 MeV are produced by the nat C(d,n) reaction using 40 MeV, 5 mA deuterons at SPIRAL2 in GANIL in France. 27) The fluxes are compared with the thermal neutron flux of the reactor at Oak Ridge National Laboratory, having a factor # 10 15 n/cm 2 /s. 28) The accelerator neutrons are characterized to be emitted in the forward-direction with respect to the deuteron beam direction. Hence, most of the emitted neutrons will be used effectively to produce 99 Mo by placing a sample after a neutron target (discussed later) in the direction of the deuteron beam. Thirdly, a quantity of 100 Mo samples of over 100 g weight mass can be used, because the neutron has no charge, and therefore the traveling range in a sample is much longer than that of a charged particle. When one uses a proton beam to produce 99 Mo or 99m Tc via the 100 Mo(p,pn) 99 Mo reaction or the 100 Mo(p,2n) 99m Tc reaction, the quantity of the 100 Mo sample mass would be less than 1 g, owing to the short proton range in the 100 Mo sample. In addition, proton beams have a heat problem of the sample because protons are stopped inside a 100 Mo sample material, and therefore high-intensity proton beams of above a few hundred µA are hardly used to produce medical radioisotopes.

99 Mo yield.
In obtaining the specific activity of the produced 99 Mo as high as possible we calculated the angular and thickness distributions of the yield of 99 Mo produced by irradiating a 100 Mo sample having a large surface area with accelerator neutrons provided by the nat C(d,n) reaction at a deuteron energy (E d ) of 40 MeV. 29) In the calculation we assumed that the 100 Mo sample would be placed 2 cm downward from the carbon target, as shown in Fig. 5a, and the 40 MeV deuteron beam size to be a point. We used the latest data of neutrons from the nat C(d,n) reaction at E d F 40 MeV 30) and the eval-uated cross section of 100 Mo(n,2n) 99 Mo given in the Japanese Evaluated Nuclear Data Library (JENDL-4.0). 26) As we can see in Fig. 5b, the calculated 99 Mo yield is mostly distributed in a narrow region at an extremely forward angle with respect to the deuteron beam direction and within a sample thickness of as thick as about 4 cm. From this result, an appropriate shape of the 100 Mo sample to obtain a high specific activity of 99 Mo is determined to be cylindrical. Next, we measured the yield of 99 Mo produced by the nat Mo(n,2n) 99 Mo reaction to make a rigorous test of the measured energy and angular distributions of the accelerator neutrons, including the evaluated cross section. 31 Mo,9.25% for 94 Mo,15.92% for 95 Mo,16.68% for 96 Mo,9.55% for 97 Mo,24.13% for 98 Mo and 9.63% for 100 Mo. 33) The accelerator neutrons   were provided by the nat C(d,n) reaction using 40 MeV deuterons at the azimuthally variable field (AVF) cyclotron at Cyclotron and Radioisotope Center (CYRIC), Tohoku University. 34) The distance, d, between the carbon target and the nat MoO 3 sample was 9 mm. The activity of 99 Mo at the end of irradiation (EOI) was determined, as given in Table 4, by considering the branching ratio of the observed .rays and the .-ray detection efficiency of the HPGe detector. The self-absorption of the .-rays in the irradiated 100 MoO 3 sample was corrected by using a photon cross-sectional database provided by the National Institute of Standards and Technology. 35) Next, we estimated the yield of 99 Mo by numerical calculations for a comparison with the measured yield, as follows. We used the neutron-nucleus reaction cross sections given in the fourth version of the Japanese Evaluated Nuclear Data Library (JENDL-4.0) for molybdenum and oxygen in the 100 MoO 3 sample. In calculating the neutron flux we used the latest data of the cross section, which was obtained by irradiating a 15-mm-thick carbon target with 40 MeV deuteron beams. 30) We corrected for the difference in the attenuation of the neutron flux inside the carbon target because the thickness of the carbon target used in the aforementioned study 30) was 15 mm, compared with 10 mm in this measurement. The correction was made by using the simulation code Particle and Heavy Ion Transport code System (PHITS). 36) The estimated yield of 99 Mo for each set of the nat Mo samples is in good agreement with the measured yield, as given in Table 4, which reflects the accuracies of the measured neutron energy and the angular distributions of the neutrons from the nat C(d,n) reaction and the accuracy of the evaluated cross section of the 100 Mo(n,2n) 99 Mo reaction.
Based on the good agreement between the measured 99 Mo yield and the calculated yield, we calculated the activities of 99 Mo at the EOI, produced by the 100 Mo(n,2n) 99 Mo reaction for an enriched 100 MoO 3 sample, in terms of the weight and radius (r s ) of the 100 MoO 3 sample, the distance (d) between the carbon target and the 100 MoO 3 sample, and the irradiation time (i t ). 31) In the calculation, we used the latest data on the angular and energy distributions of neutrons from nat C(d,n) at E d F 40 MeV (assuming a beam intensity of 2 mA) and evaluated cross sections given in JENDL-4.0. 26) Some of the calculated 99 Mo yields are listed in Table 5. Here, we took the radius of the deuteron beam, r d , of 0.5 cm and i t F 24 h as a typical setup to reduce the heat power density in the carbon target deposited by the deuteron beam and the decay loss of 99 Mo produced during the neutron irradiation period. We can see that both the 99 Mo and 99m Tc yields have a maximum at r s F 2.0 cm independent of d in most cases. 99 Mo/ 99m Tc in Japan. In order to calculate the activity of 99m Tc, which is obtained daily using 99 Mo produced every day by the 100 Mo(n,2n) 99 Mo reaction, we first calculated the 99 Mo activity reserved daily. Secondly, we compared the calculated activity with the current demand of 99 Mo, which is estimated by considering the number of diagnostic procedures by using 99m Tcradiopharmaceuticals every year in Japan. 31) As given in Table 5 we calculated that a certain amount of 99 Mo is produced every day by the 100 Mo(n,2n) 99 Mo reaction with 40 MeV, 2 mA deuteron beams (24 h irradiation). Through our studies discussed later we consider that an enriched 100 MoO 3 sample of mass 150 g (100 g of 100 Mo) is one of the feasible cases in terms of the present production efficiency of 99 Mo, and the elution performance of 99m Tc from 99 Mo, and the shortest distance between the carbon target and the 100 MoO 3 sample will be practically d F 1.0 cm by taking account of the mechanical structure of a rotating carbon target system (discussed later). Since we obtain the maximum (calculated) yield of 99 Mo for a sample radius of r F 2.0 cm for most cases (Table 5), the calculated yield of 99 Mo at the EOI per day is 657 GBq (150 g 100 MoO 3 sample mass, r s F 2.0 cm, and d F 1.0 cm). The daily production of 657 GBq (18 Ci) of 99 Mo provides 2.30 TBq (63 Ci) of 99 Mo on average after 10 days in steady-state operation of the cyclotron, as shown in Fig. 7. It should be mentioned that we can also obtain the 99 Mo yield of 700 GBq using a 100 MoO 3 sample of 100 g by employing a following neutron irradiation scheme of the sample. The calculated yields of 99 Mo for one-pellet (100 g) and two-pellet (2 # 100 g) 100 MoO 3 samples are 505 and 756 GBq (Table 5), respectively. Hence, the yield

Capability to meet demand for
where 6 1 F 0.0105/h and 6 2 F 0.1155/h are the decay constants for 99 Mo and 99m Tc, respectively, and (A Mo ) 0 is the activity of 99 Mo at t F 0.   100 Mo) for 24 h in terms of the radius and thickness of 100 Mo, and the distance (d) between the natural carbon target and the 100 Next, the 99m Tc activity that might be obtained daily at the end of thermochromatographic separation, discussed later, is evaluated. After the EOI, the irradiated 100 MoO 3 sample will be placed in a thermochromatography apparatus to separate 99m Tc from the neutron irradiated sample containing 99 Mo. Taking into account the decay loss of 99m Tc during the separation procedure for about 2 h and the separation efficiency of 80%, the 99m Tc activity immediately after separating 99m Tc from the "old" and "new" 99 Mo is 1.30 TBq (35 Ci). Here, "old" and "new" 99 Mo indicate the activity of 99 Mo produced on previous days and that day, respectively. Note that a 99m Tc solution is usually eluted twice per day, and therefore the obtained 99m Tc activity is 2.21 TBq (60 Ci), 1.7-times that in the case of elution once a day (see Fig. 8).
We next discuss the demand of 99 Mo in Japan by considering the number of diagnostic procedures currently carried out by using 99m Tc radiopharmaceuticals. About 0.9 million diagnostic procedures (2,750 procedures/day) have been performed every year using 99m Tc radiopharmaceuticals with an average dose of about 740 MBq (20 mCi) at the time of injection to a patient: 2.05 TBq (55 Ci) of 99m Tc is used every day. The activity of 99m Tc needed in Japan every day at a radiopharmaceutical company to prepare 99m Tc radiopharmaceuticals is calculated to be 4.10 TBq (110 Ci) by referring to the reports by Pillai et al., 38) Bennett et al., 39) and Ross et al. 40) as follows. In the U.S.A., about 50,000 diagnostic procedures are carried out daily using 99m Tc radio-pharmaceuticals. The U.S.A. requires 55.5 TBq (1,500 Ci) of 99m Tc daily assuming that 1.1 GBq (30 mCi) of 99m Tc is injected into a patient, and by considering the decay loss during the transportation of 99m Tc from the radiopharmaceutical company to hospitals, a total 99m Tc activity of 111 TBq (3,000 Ci) is required every day. Similarly, the total 99m Tc activity required in Japan is calculated to be 4.10 TBq (110 Ci) every day.
As discussed above, 2.21 TBq (60 Ci) of 99m Tc is obtained at the end of the separation of 99m Tc, i.e., immediately before forming the 99m Tc radiopharmaceuticals at a radiopharmaceutical company. When one can inject the 99m Tc radiopharmaceuticals prepared by using the 2.21 TBq 99m Tc within 6 h to a patient, about 50% of the daily procedures using 99m Tc radiopharmaceuticals can be performed in Japan. In fact, because about 35% of Japan's population live in the capital-area (a region within about 150 km radius from the center of Tokyo, e.g., Kanagawa, Saitama, Chiba, Ibaraki, Tochigi, Gunma, and Yamanashi prefectures), and the number of people who live in the capital-area, Kansai-area, and Chukyo-area is about 60% of the total population of Japan, 99m Tc radiopharmaceuticals can be delivered from a radiopharmaceutical company to those who live there within 6 h. The proposed delivery of 99m Tc would be possible by using the current delivery system of 18 F-FDG, 41) a radiopharmaceutical fluorodeoxyglucose containing 18 F with a half-life of 1.8 h shorter than 99m Tc (T 1/2 F 6 h). In Japan, deliveries of 18 F-FDG are being carried out three times per day by road transport from the 18 F-FDG-producing radiopharmaceutical company to hospitals about 200 km away within 3 h. The activity of 18 F prepared at the radiopharmaceutical company is three-times stronger than that needed for injection into a patient at a hospital by considering a decay loss of 68% during the transport of 18 F in 3 h. Note that the 99m Tc activity needed at hospitals for the 99m Tc procedures to satisfy 50% of the demand in Japan is 10.2 TBq (27.5 Ci). The decay loss of 99m Tc activity in transportation for 3 h is 30%. Hence, the 99m Tc activity of 60 Ci obtained at a radiopharmaceutical company would be enough to perform the mentioned procedures. Note that we decided to harvest the produced 99 Mo every day by considering the 100 Mo inventory and the 99 Mo decay during the 100 Mo sample irradiation. A harvest frequency of 99 Mo once every six days would decrease the instantaneous production rate by about 40% relative to that of one day. 3.1.4. Radionuclide purity of 99 Mo. The radionuclide purity of 99 Mo should be high in order to obtain high-quality 99m Tc to perform the separation process of 99m Tc from 99 Mo under radionuclides with little impurities, and not to create a problem for the storage of long-lived radioactivity. We measured the radionuclide purity of 99 Mo produced by using an enriched 100  The activities of 99 Mo and 97 Zr at the EOI were determined to be (3.16 ' 0.12) # 10 6 Bq for 99 Mo and (31.5 ' 1.6) # 10 3 Bq for 97 Zr, which is 1% of the 99 Mo activity, as shown in Table 6. Namely, 99 Mo was produced with a minimum level of radioactive waste and without radioisotopes of Tc other than 99m Tc and 99 Tc (T 1/2 F 2.1 # 10 5 y). They are important because the irradiated enriched 100 MoO 3 sample can be recycled.
3.2.1. Medical radioisotopes produced in reactors and accelerators. In the treatment for patients with cancers, medical RIs are used first to obtain pretherapy imaging information concerning biodistribution and dosimetry in patients, and second to perform higher dose targeted molecular therapy in the same patients. Most medical RIs used for imaging and therapy are, respectively, being produced in accelerators (except 99 Mo) and in reactors (except 225 Ac). In the production of therapeutic RIs in reactors by the fission reaction of 235 U or the thermal neutron capture reaction of a sample, high thermal neutron fluxes on the order of 10 14 n/(cm 2 s) and the use of a large quantity of a sample plays a key role. Note that carrier-free (without any isotope of a sample nuclide of 235 U) RIs suitable for medical use are obtained by the fission reaction. Currently, carrier-free RIs of 90 Y (T 1/2 F 2.67 d), the daughter radioisotope of 90 Sr (T 1/2 F 28.8 y), and 131 I (T 1/2 F 8.02 d) have been used for therapy. On the other hand, carrier-added (with a sample) RIs are usually generated by the thermal-neutron capture reaction. They cannot be separated from a neutron-irradiated sample because the atomic number of RIs is the same as that of a sample. However, there are several cases  (keV) Lu reactions. Here, it must be noted that currently constantly available RIs produced by the fission reaction and thermal neutron capture reaction are, 89 Sr, 90 Y, 131 I, 177 Lu, 192 Ir, and 198 Au etc. We might expect that a wide variety of medical radioisotopes can be produced in reactors by the two reactions mentioned above. A limitation of available numbers comes mainly from the fission yield curve of 235 U having maxima at masses of around 90-100 and 133-143, as shown in Fig. 10, and a sample mass dependence of the thermal neutron capture reaction having a large cross section. 43) In accelerators a wide variety of carrier-free RI with a high specific activity, mostly used for diagnostics, have been produced by using proton beams. In proton irradiation on a sample, it must be noted that the whole proton energy is transformed into heat in the sample, and the traveling range of protons in a sample is much shorter than that of neutrons, which limit both the proton beam intensity and the quantity of the sample for producing RIs. Hence, therapeutic RIs are hardly produced using proton beams. We first proposed a new route to produce 99 Mo by using accelerator neutrons, and then proposed new methods to produce therapeutic RIs. 44)- 46) 3.2.2. Production for theranostic radioisotopes. A charge-exchange reaction, such as (n,p), (n,x), and (n,,), of a sample nucleus with a medium-weight mass, has a sizable cross section of from 950 to 9500 mb at a neutron energy of between 910 and 930 MeV, which is almost independent of the mass number of the sample. Here, (n,x) denotes the (n,nBp) and (n,d) reactions. The cross section of the (n,2n) reaction of a neutron-rich nucleus at 910 < E n < 20 MeV is also quite large, and does not depend so much on the nuclear mass; it is in the range between 500 and 2,000 mb. Note that the neutron has no charge, and therefore the traveling range in a sample is much longer than that of a charged particle. Therefore, using high flux accelerator neutrons and a large amount of a sample, a large quantity of a wide variety of carrier-added and carrier-free radioisotopes (without any other isotope of a sample nuclide) can be produced, which would lead to a new era in theranostic RIs production. A schematic view of the production of a variety of therapeutic radioisotopes using accelerators and many stable isotopes is shown in Fig. 11.
We have proposed new routes to produce carrier-free medical radioisotopes of 90 Y, 43) 64 Cu, and 67 Cu 44),45) using accelerator neutrons provided by the nat C(d,n) reaction. Successful PET with the use of 18 F for assessments of tumor characterization has triggered a search for a longer half-life PET RI to diagnose the dynamics of a medicine in a living body that has a slow reaction time. 64 Cu with a half-life of T 1/2 F 12.7 h, longer than that of 18 F (T 1/2 F 1.8 h), is considered to be a promising RI suitable for labeling many radiopharmaceuticals for PET imaging, 2) since 64 Cu decays by positron (O D ) emission. The Cu radioisotope is known to have unique potentials useful for diagnostic imaging and in targeted radionuclide therapy. 4), 5) In radioimmunotherapy (RIT) for tumor treatments, 90 Y (T 1/2 F 64 h), a pure O ! -   11. Schematic view of the production of a wide variety of therapeutic radioisotopes. A hot nucleus is produced by bombarding a sample with accelerator neutrons, followed proton, neutron or ,-particle emission. When a charged particle, such as a proton or ,, is emitted, one can obtain carrier-free RIs by employing a chemical separation technique.
ray emitter with an average O ! -ray energy of 935 keV, is most widely used to kill large tumor masses, since the range of O ! -rays in H 2 O is as long as 12 mm. 67 Cu (T 1/2 F 62 h), a pair radioisotope of 64 Cu, is considered to be a promising radionuclide for treating small distant metastases of up to 4 mm in size in radioimmunotherapy. 6) 67 Cu has unique nuclear properties and chemical behavior for use in RIT. 4),5) Namely, 67 Cu emits O ! -rays with an average energy of 141 keV, which allow radiopharmaceuticals of 67 Cu to provide a lethal dose of radiation to target cancer cells. 67 Cu also emits 185 keV .-rays, which permit SPECT imaging during therapy. In addition, 67 Cu has a sufficiently long half-life (T 1/2 ) of 62 h, allowing it to be delivered to tumors, which may take 24 to 48 h to reach their peak concentration in tumors. 4) 64 Cu should be noted to be used for pretherapeutic PET studies for accurate evaluations of the dose delivered to a normal organ before the injection of 67 Cu for RIT to patients, since the irradiation of vital organs should be minimized. The coordination chemistry of copper applied to the production of radiopharmaceuticals has been well established. On the basis of a successful clinical study on a radiopharmaceutical containing 67 Cu for B-cell non-Hodgkin lymphoma, about 450 TBq (12,000 Ci) of 67 Cu is considered to be required per year in the U.S.A. 47) Currently, there exists no technology to meet such a demand.
Thus far, many studies have been carried out to produce 64 Cu and 67 Cu in reactors or accelerators. 64 Cu was produced by the 64 Zn(n,p) 64 Cu reaction in reactors using accelerators by the 64 Ni(p,n) 64 Cu, 64 Ni(d,2n) 64 Cu, 64 Zn(d,2p) 64 Cu, 66 Zn(d,,) 64 Cu, 68 Zn(p,,n) 64 Cu, and 64 Zn(n,p) 64 Cu reactions in accelerators. 47) Among the studies, the generally adopted production route is the 64 Ni(p,n) 64 Cu reaction, which provides a high specific activity 64 Cu using a highly enriched 64 Ni target. The maximum production yield of 64 Cu is expected to be about 37 GBq (1 Ci) by bombarding a 64 Ni sample with 50 µA proton beams for 12 h. 48) It is very greatly encouraged to increase the availability of 64 Cu. 67 Cu has been produced by the 67 Zn(n,p) 67 Cu reaction in both reactors and accelerators, 13) and in accelerators via the 68 Zn(p,2p) 67 67 Cu reactions are currently used for the production of 67 Cu. Since the proton energy used in the 68 Zn(p,2p) 67 Cu reaction is high, a large amount of impurity RI of 64 Cu is produced by the 68 Zn(p,,n) 64 Cu reaction at EOI. Regarding the 67 Cu production induced by accelerator neutrons, the spectrum-averaged cross section of the 67 Zn(n,p) 67 Cu reaction was measured at E n F 4.95 MeV, 49) but the route has not yet been adopted because of the small cross section and the lack of an intense neutron source. Note that recently high-quality 67 Cu has been produced by the 68 Zn(.,p) 67 Cu reaction at Argonne National Laboratory. 50) Concerning the 64 Cu and 67 Cu productions using accelerator neutrons, the main drawback comes from the low neutron flux, but not from the nuclear reaction processes, such as the cross section of a required RI or the high production yield of an impurity RI. As mentioned above a high neutron flux of 910 15 n/s at an average neutron energy of E n : 14 MeV can be obtained owing to recent progress in accelerator technology. These findings led us to propose new routes to produce carrier-free radioisotopes of 64 Cu by the 64 Zn(n,p) 64 Cu reaction and 67 Cu via the 67 Zn(n,p) 67 Cu and 68 Zn(n,x) 67 Cu reactions using accelerator neutrons. In order to calculate the production yields of 64 Cu and 67 Cu we first re-measured the cross sections of the 64 Zn(n,p) 64 Cu, 67 Zn(n,p) 67 Cu, and 68 Zn(n,x) 67 Cu reactions using neutrons at E n : 14 MeV. Although many studies were carried out to measure those cross sections at E n : 14 MeV, there remained significant differences between different data sets. The measurement was carried out using 914 MeV neutrons produced via the 3 H(d,n) 4 He reaction at the Fusion Neutronics Source (FNS) facility of Japan Atomic Energy Agency (JAEA). 51) The obtained results led us to estimate the yields of 64 Cu produced by the 64 Zn(n,p) 64 Cu reaction and 67 Cu by the 67 Zn(n,p) 67 Cu and 68 Zn(n,x) 67 Cu reactions using high neutron fluxes. The neutrons could be provided by the nat C(d,n) reaction with 40 MeV, 5 mA deuteron beams. 27) The estimation was performed using the evaluated cross section of the neutron induced reaction on Zn isotopes given in the Japanese Evaluated Nuclear Data Library. The yield of 64 Cu was calculated to be 1.8 TBq/(175 g of 100%-enriched sample of 64 Zn) for an irradiation time of 12 h, which is much larger than the expected 64 Cu yield of 37 GBq via 64 Ni(p,n) 64 Cu, and that of 67 Cu via the 68 Zn(n,x) 67 Cu reaction was calculated to be 287 GBq/(186 g of 100%-enriched sample of 68 Zn) at EOI for an irradiation time of 2 days, 45) which is much larger than a reported yield of 10 GBq by  65 Zn (T 1/2 F 244 d at 1,116 keV), and 69m Zn (T 1/2 F 13.8 h at 439 keV). The isotope assignments of the .-rays were made on the basis of their energies and decay curves. 65 Zn was identified by the 1,116 keV .-ray.
The radionuclide purity of 67 Cu was determined to be extremely low compared with those produced by the 68 Zn(p,2p) 67 Cu and 70 Zn(d,,n) 67 Cu reactions, as given in Table 7, and estimated ones to investigate a possible reaction for producing impurity radionuclides. The estimation was made using the isotope composition of the enriched 68 ZnO sample mentioned above, the neutron energy spectra from the nat C(d,n) reaction, and neutron nuclear reaction cross-sectional data on Zn isotopes. 26) The estimated activity ratios of the impurity radioisotopes agree with the experimental ratios within the experimental uncertainties, as given in Table 7. 53) The obtained information is also important when purchasing an expensive sample with a variety of isotopic compositions.
3.3. Production of radioisotopes in polyethylene blocks. We have proposed another production route by using accelerator neutrons backscattered by materials, such as polyethylene or lead blocks.
3.3.1. Experiment for producing RIs using polyethylene blocks. In studying the production routes of medical RIs with accelerator neutrons by using a sample that was covered with polyethylene blocks to reduce the neutron background in an experimental room, we happened to find much larger yields of some of the RIs than those without polyethylene blocks. 54) This study was performed by irradiating five stacked samples of 93 Nb, enriched 68 ZnO, enriched 64 ZnO, natural nat ZnO, and enriched 90 ZrO 2 as well as two stacked samples of 93 Nb and enriched metallic  Zn with accelerator neutrons. The 93 Nb sample was used as a high-energy neutron-fluence monitor. 54) The masses for the 68 ZnO, 68 Zn, and 64 ZnO samples were about 360 mg, and the enrichment of the 68 ZnO, 68 Zn, and 64 ZnO samples was over 99%. These samples were covered with polyethylene (or lead) blocks, as shown in Fig. 13a. The size of the individual polyethylene or lead block was 200 # 100 # 50 mm 3 . The distance (d) between the Al holder and the polyethylene (or lead) shown in Fig. 13b was set to be either 3 or 6 cm for investigating a possible effect of the blocks on the yields of produced RIs. The neutrons were provided from the deuteron breakup reaction on a 9 Be target using a 50 MeV, 0.5 µA deuteron beam at TIARA-QST. The samples were irradiated for about 15 min.
Hereafter we focus on the results of the two samples, 68 ZnO and 68 Zn; 68 Zn(PE) and 68 Zn(Pb) indicate 68 Zn samples covered with polyethylene and with lead, and 68 Zn(no) stands for a 68 Zn sample with neither polyethylene nor lead, respectively. Typical .-ray spectra of the irradiated 68 ZnO(no), 68 ZnO(PE), and 68 Zn(PE) samples placed at d F 3 cm are shown in Figs. 14a, 14b, and 14c, respectively. Identifications of the produced radioisotopes were made based on the .-ray energies and/or the absolute .-ray branching ratio (I . ), as given in Table 8.
An anomalous nuclear reaction phenomenon was found in this study. Namely, significant amounts of proton-induced reaction products of 66 Fig. 14a. We also found that the .-ray intensities    Fig. 14c were much smaller than those of the oxide 68 ZnO(PE) sample in Fig. 14b. However, the .-ray intensities of 67 Cu, 65 Ni, and 65 Zn of the 68 ZnO(PE) and 68 Zn(PE) samples were approximately the same as those of the 68 ZnO(no) sample.
The activities for various isotopes of the 68 ZnO and 68 Zn samples at the end of irradiation (EOI) were obtained, as shown in Table 9, where the corresponding nuclear reaction path and the reaction threshold energies are also indicated. Here, as an example, in order to know the dependence of the activities for the 68 ZnO(PE) sample on d we took the difference between the activities for the 68 ZnO(PE) and for the 68 ZnO(no) samples and divided this difference by the activity for 68 ZnO(no) (columns F and G for d F 3 and 6 cm, respectively, in Table 9). The enhancement factors for d F 3 and 6 cm are compared by taking their ratio (column H in Table 9).
The results given in Table 9 are summarized in terms of the activity A(X) of a particular radioisotope X at the end of irradiation (EOI), as follows. Based on these findings we consider that they indicate a main reaction process to generate protons and neutrons, which play a key role in the large productions of 67 Ga, 66 Ga, 64 Cu, and 69m Zn. The second part of the summary, their smaller yields at d F 6 cm, suggests that they might be generated by any interaction between primary neutrons (hereafter n prim ) produced by the Be(d,n) reaction and nuclei in the polyethylene blocks. On the other hand, the third part of the summary indicates that these protons and neutrons should be dominantly generated in the 68 ZnO(PE) sample not in the polyethylene blocks via some type of nuclear interaction between neutrons scattered backwards by the polyethylene blocks (hereafter n sc ) and oxygen nuclei in the 68 ZnO sample. The anomalous large yields are considered to be dominated by n sc . This is apparently unexpected, since the flux of n sc is much reduced compared with that of n prim , and the energy of n sc is lower than that of n prim . When the energy of n sc is low, that of protons produced by any reaction process between n sc and nuclei in the samples is also low. The cross sections of the 68 Zn(p,2n) 67 Ga reaction with a threshold energy of 12.2 MeV and the 68 Zn(p,3n) 66 Ga reaction with a threshold energy of 23.6 MeV decrease with a proton energy of less than about 26 MeV. 55) Next, we studied the neutron energy dependence of anomalous large yields observed at E d F 50 MeV by using 40 MeV deuterons. Enriched 68 ZnO samples with and without polyethylene blocks were irradiated with 40 MeV deuterons. Note that the neutron energy from 40 MeV deuterons is smaller than that from 50 MeV deuterons. The measured activities of 67 Cu, 69m Zn, and 64 Cu for the 68 ZnO sample are given in Table 10. Contrary to the results for 50 MeV deuterons, the yields of the observed radioisotopes for 40 MeV deuterons were independent of the existence of polyethylene blocks. It is evident that the primaryneutron energy played a key role in the large yields of the various radioisotopes mentioned above.

Estimation with Particle and Heavy Ion
Transport code System (PHITS). We calculated the produced activities, A(X), for 68 ZnO(no) and 68 ZnO(PE) samples for 50 and 40 MeV deuterons to compare with the measured activities. The calculation was performed using the PHITS code with the geometry of the experimental setup shown in Fig. 13, and the evaluated production cross section data of neutron-and proton-induced reactions given in the fourth version of the Japanese Evaluated Nuclear Data Library (JENDL-4.0/HE). The neutron production rates from the 9 Be(d,n) reaction at E d F 50 and 40 MeV were determined so that the PHITS simulation gives the same angular and energy distributions of neutrons as those of the latest data reported by Meulders et al. 56) and Saltmarsh et al. 57) Note that the neutron data at E d F 50 MeV and at E d F 40 MeV were obtained by using 10-mm-thick and 6.3-mm-thick Be targets, respectively. We used a 9-mm-thick Be target for 50 and 40 MeV deuterons. Hence, we corrected for any difference in the attenuation of the neutron fluence inside the Be target between the previous studies and the present study by using the PHITS code.
As given in Table 10, the calculated activities of 67 Cu, 67 Ga, 66 Ga, 69m Zn, and 64 Cu from the 68 ZnO(no) sample for 50 and 40 MeV deuterons are consistent with the measured ones within a factor of 3. Here, 67 Cu are produced by the 68 Zn(n,x) reaction. 67 Ga and 66 Ga are generated by the 68 Zn(p,2n) and 68 Zn(p,3n) reactions in which protons that passed the Be target after the deuteron breakup reaction on Be are considered to be bombarded on the 68 ZnO(no) sample. The 68 Zn(p,n,) 64 Cu and 64 Zn(n,p) 64 Cu reactions contribute to generate 64 Cu. Note that the isotopic composition of 64 Zn in the 68 ZnO sample is 0.047 at.%. Contrary to the cases for the 68 ZnO(no) sample, the calculated activities of 67 Ga, 66 Ga, and 69m Zn were smaller by a factor of about 15 to 75 than the measured ones, as given in Table 11, and almost the same as those obtained from the 68 ZnO(no) sample. The reason why the measured activities differ from the calculated ones is unclear at present. Some sort of nuclear reaction process between oxygen nuclei in the 68 ZnO sample and the scattered neutrons could be the origins of the discrepancy.
In conclusion we discovered anomalously large yields of 67 Ga, 66 Ga, 69m Zn, and 64 Cu by irradiating a 68 ZnO sample that was covered with polyethylene blocks with neutrons provided from the 9 Be(d,n) reaction by 50 MeV deuterons, but not by 40 MeV deuterons. This finding will be important for the production of radioisotopes in large quantity with accelerator neutrons by using not only incident accelerator neutrons on a sample, but also neutrons scattered backward from polyethylene blocks simultaneously. Especially, we could produce several radioisotopes which are normally generated by proton-induced reaction from a single sample irradiated with the backscattered neutrons. The specific activity of fission-99 Mo is high, 9370 TBq/ (g Mo), 37) and therefore 99m Tc with high specific activity that is sufficient for performing medical Tc generator. In fact, about 1 ml of saline is enough to recover 99m Tc with an activity of 740 MBq. On the other hand, a typical specific activity of 99 Mo produced by any alternative production method of 99 Mo without the fission reaction of 235 U is very low, about 1/5,000 of fission-99 Mo. 22) Hence, when one uses the 99 Mo/ 99m Tc generator, about 5 l of saline is required to collect 99m Tc of 740 MBq, which would result in an unacceptably low concentration of eluate for the direct formulation of radiopharmaceuticals. Note that a typical activity of a solution of 99m Tc radiopharmaceuticals administered to a patient is high, about 740 MBq/(a few ml). So far, in order to obtain a high specific activity, 99m Tc, from 99 Mo with such a low specific activity various method, such as chromatographic, 58) solvent extraction, 59) and thermo-separation methods, 13) have been developed.
Among these methods, upon considering the following potential we employed the thermo-separation procedure, which utilizes the different volatility of technetium oxide and MoO 3 to separate 99m Tc from 99 Mo. Namely, with this method it is expected that a large quantity of an enriched 100 MoO 3 sample can be used. 99m Tc with a high radioactive concentration can be obtained free from any chemical impurities, and an irradiated 100 MoO 3 sample can be recovered with high efficiency for recycling. So far, on the basis of thermo-separation, the procedures of sublimation and thermo-chromatography have been developed. 60) In sublimation, 99m Tc produced in a MoO 3 powder sample is separated, since it volatilizes at a temperature much lower than the melting point of MoO 3 at 795°C. In thermo-chromatography, Tc and Mo oxides volatilized from a molten MoO 3 sample are condensed in different temperature zones along a column where the temperature gradient is kept constant in a furnace (see Fig. 15). In reality, before we started to develop a thermoseparation system, the aforementioned potentials of the thermoseparation method had not yet been materialized, owing to the challenges discussed below. In fact, it was not even clear which is better for separating 99m Tc from an irradiated Mo sample of over 100 g, a sublimation method or a thermo-chromatography   61) Using the D at 780°C and Brownian motion theory, the mean diffusion distance, x, which is given as x F (2Dt) 1/2 , was calculated to be 910 !4 cm for one hour of diffusion (t F 1 h). Since D is so low, we understood that it is impossible to obtain a high separation efficiency of 99m Tc from a thick MoO 3 powder sample which is used for 99 Mo production by the 100 Mo(n,2n) 99 Mo reaction. 62) Hence, we employed a thermochromatographic separation method. 62)-64) A schematic of the experimental setup of the thermochromatographic separation is shown in Fig. 15. So far, by using the thermochromatographic separation technique many studies were undertaken to measure the separation efficiency by using a molten MoO 3 sample. 37),65),66) However, there remain several challenging problems concerning separation. First is that the separation efficiency of 99m Tc, C sp , which is the proportion of 99m Tc separated during the thermochromatographic-separation process, is low. This diminishes markedly with repeated sublimation tests (repeated milking tests) at a constant furnace temperature, T fur , and decreases with an increasing mass of MoO 3 loaded into a sublimation furnace at a time. For example, the efficiency at furnace loading of a 200 g Mo sample generated via the 98 Mo(n,.) 99 Mo reaction was 25% on average, and rarely exceeded 50%. 37) Note that a milking process of 99m Tc from 99 Mo is usually performed twice per day for at least one week. Typical separation efficiencies of a 5 g MoO 3 sample were 85, 12, 19 and 24%, 65) which were markedly diminished with repeated milking tests. An Idaho group 66) used a molten MoO 3 sample with a thickness of 0.8 mm in order to obtain a high separation efficiency of 99 Tc, a pure O-emitter.
Having such a thin sample is, however, not desirable for the large-scale production of 99 Mo. We have challenged these problems by measuring the diffusion coefficient and separation efficiency of 99m Tc from a molten MoO 3 sample of two different thicknesses using a home-made electric furnace, shown in Fig. 15. 62), 63) A set of three-stage quartz tubes was enclosed by a four-zone tube furnace. A platinum boat was used to hold the irradiated MoO 3 sample in a hightemperature region throughout the milking. The first two zones were heating sections used to melt the irradiated MoO 3 sample in a stream of oxygen carrier gas at around 830°C so as to form gaseous materials containing vaporized 99m Tc and Mo oxides. The third zone was an intermediate section used to condense any vaporized MoO 3 as a needle crystal and to transfer gaseous materials containing vaporized 99m Tc oxide from the heating sections to a final section. The fourth zone was a final section to collect the separated 99m Tc. Since the temperature of the intermediate section was set so as to decrease gradually as the distance in the intermediate section from the heating section increases, the temperature of the gaseous materials at the final section became sufficiently low so as to produce condensation of the gaseous 99m Tc products. A quartz wool filter was placed within the tube at a temperature below the melting point of MoO 3 so as to stop the migration of any volatilized MoO 3 towards the final section. Crumpled gold wire was placed in the tube to increase the surface area as much as possible for 99m Tc collection. The distribution of the 99m Tc activity along the tube was investigated using the cadmium zinc telluride (CZT) detector; its peak activity was in the condense region, clearly separated from the 100 MoO 3 sample, as shown in Fig. 16.
A sequential milking process used to separate 99m Tc from molten 99 MoO 3 samples of 4.0 and 8.5 mm thicknesses was carried out every for 924 hours at a Tc that thermally diffused through the quartz tube was confirmed by CZT-2. It should be noted that all of the 141 keV .-ray yields of a 4.0 mm (8.5 mm) thick sample obtained by CZT-1 showed the same dependence on the temperature and time, which indicates that the high separation efficiencies of 99m Tc remained constant during the sequential milking processes, which differed from the previously reported results. The 141 keV .-ray was observed after the release of 99m Tc from the molten MoO 3 sample, which included .-ray emission from the 141 keV state of 99 Tc fed by .-decay from the 181 keV state, which was populated by the decay of 99 Mo, not via the 143 keV state ( 99m Tc) (see Fig. 1), as well as the remaining 99m Tc in the sample.
The separation efficiency of 99m Tc, C sp , was derived by comparing the 141 keV .-ray intensity obtained by the CZT-1 detector after separation, Y sep , with that before the separation Y unt . Y sep and Y unt are given as the sum of the intensities of the 141 keV .-ray emitted from decays of the 143 keV state ( 99m Tc), Y(Tc), and the 181 keV state: Y(Mo). Using the Bateman equation for the parent-daughter decay, Y(Tc) and Y(Mo) are given as follows:  disintegration of 99m Tc and the total internal conversion coefficient of the 141 keV transition. Hence, the ratio R of Y sep to Y unt is given as follows: :0105tÞ Â 0:047 þ ð1 À "sepÞ Â 0:85 Â fexpðÀ0:0105tÞ À expðÀ0:1155tÞg ½expðÀ0:0105tÞ Â 0:047 þ 0:85 Â fexpðÀ0:0105tÞ À expðÀ0:1155tÞg : ½6 Using Eqs. [2] to [6], high separation efficiencies of close to 90 and 70% on average were obtained for molten MoO 3 samples of 4.0 and 8.5 mm thickness, respectively, in repeated milking processes at T fur F 845°C for t heat F 15 min, as given in Table 12. 62) The diffusion coefficient, D, of 99m Tc within the molten MoO 3 samples was derived as follows. So far, the D value of a radioactive ion in a molten sample had not yet been obtained by a release measurement of the ion. In this study, by referring to the work on foil targets for the production of radioactive ion beams at the Isotope Separator On-Line (ISOLDE) at CERN, 67) we divided the release process of 99m Tc into the following two steps: 1) the transport of 99m Tc to the surface of the molten MoO 3 sample, and 2) the evaporation of 99m Tc from the surface of the molten MoO 3 sample. In Ref. 18 the D of radioactive ions in metal targets, which were heated to high temperatures, was determined by assuming that step 2) can be neglected. Based on the same assumption, and by referring to work on the self-diffusion of radioactive ions in a sodium tungstate solution using 187 W (T 1/2 F 24 h) with a capillary method, 68) in which the evaporation process of 187 W was not included, the diffusion coefficient of 99m Tc could be derived. A fraction of the original amount of 187 W, which is left in a capillary cell at the end of the diffusion, ', is given using Fick's law of diffusion: where the parameter 3 is given by 3 F : 2 Dt/(4h 2 ), t is the time of diffusion, and h is the length of the capillary cell. Using Eq. [7] and neglecting the evaporation process of 99m Tc, the averaged values of D, hDi, of 99m Tc for the molten MoO 3 samples with 4.0 and 8.5 mm thicknesses could be derived, as presented in Table 12, where hDi is given by correcting the change of the sample thickness during the repeated milking process. The hDi value of about 1 # 10 !4 cm 2 /s is much larger than a reported value of 7.71 # 10 !12 cm 2 /s at 780°C for a powder sample.
In this study high separation efficiencies of about 90 and 70% were successfully obtained through a repeated milking process by the thermo-separation of 99m Tc from 10 and 14 g molten MoO 3 samples with thicknesses of 4.0 and 8.5 mm. By further developing the thermo-chromatography separation system we achieved a higher separation efficiency of over 90% for an irradiated MoO 3 sample of about 100 g. In these studies, the irradiated MoO 3 samples were melted in every milking process, and therefore the process could be performed under the same condition as that of the MoO 3 sample irrespective of the number of processes.
The diffusion coefficients of 99m Tc were found to be very large, and therefore 99m Tc could diffuse very rapidly in a thick molten sample within a reasonable heating time of 15 to 30 min, and subsequently evaporate from the sample. The present result solves the long-standing problems concerning the thermoseparation of 99m Tc from a MoO 3 sample with an increase in the sample mass or with repeated sublimation, and will bring a major breakthrough in the production of high-quality 99m Tc by using a massive 100 Mo sample.

Quality test of 99m
Tc from 99 Mo separated by thermochromatography. Here, we discuss the safety and efficacy of the "desired" 99m Tc radiopharmaceutical to assure parenteral administration to a patient. The United States Pharmacopeia (USP) contains regulatory requirements concerning the radionuclide purity and the radiochemical purity of 99m Tc and the concentration of aluminum (Al) in the 99m Tc product used to prepare radiopharmaceuticals. 69) Namely, the amount of 99 Mo and the total concentration of all other O ! and .-ray emitters in the 99m Tc product must be less than 0.015% and 0.01%, respectively. The radiochemical purity of 99m TcO 4 ! (pertechnetate) in a saline solution must be above 95% and the chemical purity of 99m Tc must be above 90%. Note that chemical impurities generate "undesired" 99m Tc compounds, such as free 99m TcO 4 ! , which does not bind to a ligand and thus do not accumulate in a targeted organ of a patient, and thus lead to an extra radiation dose to non-targeted organs of the patient and to cause serious errors in diagnosis. The Al concentration must be less than 10 ppm. Endotoxin, known to be a pyrogen, is another important item, since even small amounts of endotoxin can cause illness in humans. The USP sets the maximum endotoxin concentration limit to be 175 EU/V, where EU is endotoxin units and V is the maximum recommended total dose in milliliters (mL).
Currently, 99m TcO 4 ! , which meets the USP requirements, is obtained from a 99 Mo/ 99m Tc generator, in which the fission-99 Mo is loaded to an alumina column and 99m Tc in the form of 99m TcO 4 ! is repeatedly eluted from the column in a saline solution. Chemical impurities, which inhibit the labeling of the 99m Tc radiopharmaceutical complex, generate "undesired" 99m Tc compounds, such as free 99m TcO 4 ! and hydrolyzed-reduced 99m Tc. It should be noted that the current USP sets those pharmacopeia standards for 99m TcO 4 ! obtained from a 99 Mo/ 99m Tc generator, but there are no pharmacopeia standards for 99m TcO 4 ! , which is obtained by a 99 Mo (or 99m Tc) production method other than one with fission-99 Mo. 3) Therefore, it is important to investigate the pharmaceutical equivalence of 99m TcO 4 ! to that obtained from the alumina-based 99 Mo/ 99m Tc generator. In fact, such a study has been conducted concerning the production of 99m Tc by the 100 Mo(p,2n) 99m Tc reaction 70) and 99 Mo by the 100 Mo(.,n) 99 Mo reaction, 71) but it has not yet been performed in the production of 99m Tc by the 100 Mo(n,2n) 99 Mo reaction. Hence, we have tested the pharmaceutical equivalence of 99m TcO 4 ! obtained by 99m Tc from the thermochromatographic separation procedure to that obtained from the alumina-based 99 Mo/ 99m Tc generator. In addition, we have studied quality-control specifications associated with new variables, such as the contamination arising from an enriched 100 Mo sample. The USP does not set any criteria concerning the nonradioactive (stable) Mo content in 99m Tc-radiopharmaceuticals because an enriched 235 U sample, which does not contain a stable Mo, and the 99 Mo/ 99m Tc generator are used for the production of 99m Tc using the fission- 99 Mo. We have adopted a dosage limit of 1,700 µg/day, which is given as the injection agent of stable Mo in a report of the International Conference on Harmonization Guideline for elemental impurities (ICH Q3D), 72) which does not cover radiopharmaceuticals, but "is intended to provide guidance for registration applications on the content and qualification of impurities in new drug substances produced by chemical syntheses". We have also studied quality-control tests of 99m Tc-radiopharmaceuticals commonly used for the imaging of brain perfusion ( 99m Tc-ECD), myocardial perfusion ( 99m Tc-MIBI), and kidney ( 99m Tc-MAG3), to ensure the safe clinical use of 99m Tc obtained by the 100 Mo(n,2n) 99 Mo reaction. The separation of 99m Tc from an irradiated 100 MoO 3 sample was carried out by the thermochromatographic method, 62),63) discussed in chapter 4.1.
The quality-control tests of a 99m TcO 4 ! saline solution on the radiochemical purity and the radiochemical yields of the 99m Tc-radiopharmaceuticals, such as 99m Tc-ECD, 99m Tc-MIBI, 99m Tc-MAG3, and 99m Tc-MDP, were performed by paper chromatography and by thin-layer chromatography, respectively. The radionuclide purity was studied by taking a .-ray spectrum of the purified 99m TcO 4 ! solution using a high-purity Ge (HPGe) detector. The Al concentration of separated 99m Tc was checked by using an Al test paper. Details of preparing 99m Tc radiopharmaceuticals using commercially available labelling kits (FUJIFILM RI Pharma Co., Ltd., Japan) are given in Ref. 73. The stable Mo content in a 3 mL 99m TcO 4 ! saline solution was measured by inductively coupled plasma-atomic emission spectroscopy (ICP-AES). Tests of the endotoxin concentrations were carried out by following the statement that the pyrogenicity of a 99m TcO 4 ! solution from a particular production procedure should be verified by having a portion of it tested by an independent qualified professional using accepted procedures. The results of the quality assessments of the 99m TcO 4 ! saline solution and 99m Tc-radiopharmaceuticals were shown to satisfy the USP requirements listed in Table 13. 73) The endotoxin concentrations were below the limit of detection (0.03 EU/mL), much less than the established limit in pharmacopoeias. The measured maximum value of the stable Mo content was 0.138 ppm, i.e., 0.138 µg/mL, which is much less than the permitted daily exposure of 1,700 µg/day given in the ICH Q3D guideline. These results provide important evidence that 99m Tc prepared by thermochromatographic separation using 99 Mo produced by the 100 Mo(n,2n) 99 Mo reaction can be a promising substitute for the fission product 99 Mo.

Recovery of an irradiated 100 MoO 3 sample.
It is important to recycle an enriched (expensive) 100 Mo sample irradiated by neutrons, because a loss fraction of the quantity of the 100 Mo sample of 100 g during neutron irradiation for 24 h is estimated to be small, about 0.0001%. Here, accelerator neutrons are assumed to be produced by the nat C(d,n) reaction using 40 MeV, 2 mA deuteron beams, similarly to the estimation given in Table 5. We have therefore developed a recovery method of 100 MoO 3 having a recovery efficiency, C rec , higher than 99% to mitigate the cost of a 100 MoO 3 sample loss. 74) So far, the recovery efficiency, C rec , of the 100 MoO 3 samples in the range of 84-97% has been reported in a study of the thermochromatography of 94m Tc (T 1/2 F 52 min) from 94 MoO 3 75) ; a recovery efficiency of 87% 66) or 90% 76) was obtained in a 99 Mo or 99m Tc production study based on the 100 Mo(.,n) 99 Mo or 100 Mo(p,2n) 99m Tc reactions by employing a chemical process.
The present recovery test was performed using a home-made electric furnace, shown in Fig. 18.
An irradiated enriched 100 MoO 3 sample of 26.450 g was divided into three platinum crucibles in a vertical three-zone tubular electric furnace along with 103.253 g of non-neutron-irradiated 100 MoO 3 to bring the sample mass up to 129.703 g so as to develop a recovery method of over 100 g of 100 MoO 3 . After the milking process was carried out eight times in total, 100 MoO 3 was recovered in two batches from the first five and the following three milking processes. The recovery was focused on the two main deposition sources (the sample remaining in crucibles and needle crystals) in the first batch (run 1), and the detailed distribution of 100 MoO 3 was investigated in the second batch (run 2), including in the quartz tube, the platinum shelf, and the quartz wool. The recovery was studied gravimetrically by measuring the weight of 100 MoO 3 remaining in the crucibles and being deposited onto any quartz pieces that 100 MoO 3 could travel through. All of the crucibles and quartz pieces were weighed before and after the thermochromatography to determine the amount of recovered 100 MoO 3 free of any possible contamination caused by the process. Firstly, the 100 MoO 3 sample remaining in the crucibles was melted at 830°C in an electric furnace and collected into a quartz test tube using a funnel. Secondly, the 100 MoO 3 needle crystals were collected by washing them off the holder with pure water, and then evaporated to dryness to measure the weight. Thirdly, the quartz wool that trapped 100 MoO 3 crystals was heated to above 830°C, and thereby the 100 MoO 3 alone was separated from the quartz wool by thermochromatography. Any quartz pieces, including the funnel used to channel molten 100 MoO 3 into the quartz test tube, were washed with pure water in an ultrasonic bath to recover the most 100 MoO 3 possible in run 2. The collected 100 MoO 3 crystals were evaporated to dryness to measure the weight. The distribution and recovery yield of the 100 MoO 3 mass after thermochromatography is summarized in Table 14. Concerning run 1, after thermochromatography, 118.870 g (92%) out of the initial 100 MoO 3 mass of 129.703 g was found to remain in the crucibles, while 9.477 g (87%) out of the 10.833 g of 100 MoO 3 that was vaporized from the crucibles was trapped as needle crystals. During the recovery process of 118.870 g of 100 MoO 3 in the crucibles, a certain amount of 100 MoO 3 was vaporized and then trapped as needle crystals. Hence, the recovery from the needle crystals was 10.296 g, which was more than the needle crystal mass of 9.477 g measured after thermochromatography. Finally, 117.502 g (99%) out of 118.870 g of 100 MoO 3 was recovered from the crucibles and 10.296 g from the needle crystal holder, which gave a recovery yield of 98.5% (127.798/129.703 F 0.985). For run 2, 110.551 g (94%) out of the initial 100 MoO 3 mass of 117.490 g was found to remain in the crucibles after thermochromatography. The vaporized 100 MoO 3 from the crucibles (6.939 g) was deposited onto the quartz tubes (0.073 g), quartz wool (0.376 g), and a needle crystal holder (6.380 g), and was 6.829 g in total. The recovered 100 MoO 3 from the crucibles was 108.640 g (92%), and that from other than the crucibles was 8.426 g, giving a total recovery of 117.066 g (99.6%). Note that 1.911 g of 100 MoO 3 was vaporized from the crucibles and then trapped as needle crystals during the recovery process, which resulted in the total amount of recovered 100 MoO 3 deposited, other than in the crucibles (8.740 g) being larger than the amount of needle crystals measured after the separation.
A high recovery yield of 99% was obtained, which would significantly reduce any financial damage due to the loss of the enriched 100 MoO 3 sample. We consider that the newly developed home-made thermochromatography system should have a capability of nearly 100% recovery, because all of the 100 MoO 3 , including the small unrecovered amount, is kept within the thermochromatographic apparatus inside the electric furnace. 64 Cu and 67 Cu and biodistribution of 67 64 Cu and 67 Cu produced by the 64 Zn(n,p) 64 Cu, 67 Zn(n,p) 67 Cu and 68 Zn(n,x) 67 Cu reactions using nat ZnO or 64 ZnO samples. 77) A flowchart of the separation steps of the irradiated samples is given in Fig. 19.

Separation of
The irradiated sample of 5.225 g was dissolved in 20 ml of 36 wt% HCl, which was passed through an ion-exchange column for adsorbing the Cu ions, and thus separating Zn. The 64,67 Cu was then eluted with 20 ml of 2.0 M HCl, passed through an anionexchange column to remove traces of Zn, followed by washing with 10 ml 2.0 M HCl to obtain purified 64,67 Cu radionuclides. The collected efficiency of 64 Cu separated from the irradiated 64 ZnO sample was 96%. The time required for the column separation process was 3-4 h. We also developed a method for recycling irradiated enriched 64 Zn and 68 Zn samples after radiochemical separation. The recovery efficiency of the nat ZnO sample was demonstrated to be over 95% in a cold (non-radioactive) run using an alkaline precipitation method. The purified 64,67 Cu solution was reacted with a bifunctional ligand used for antibody labelling. The labelling yield was determined by thin-layer chromatography (TLC) to be good at 92-97% which was satisfactory for clinical radiotherapy applications.

Biodistribution of 67 CuCl 2 in tumor-bearing mice.
We developed new production routes to improve the low-availability of the promising radionuclide of 64 Cu and 67 Cu, and to establish a radiochemical method for obtaining high-quality 64 Cu and 67 Cu from neutrons irradiated Zn samples. Cu-based radiopharmaceuticals that can accumulate in cancer cells, such as 64 Cu-labeled proteins, peptides, and antibodies, have been developed and widely used. 4) However, currently 64 Cu complexes are considered to have relatively low stability in vivo, which could cause the loss of 64 Cu from the complexes, leading to less accumulation of 64 Cu in targeted cancer cells by producing free radioactive 64 Cu. 5) 64 Cu chloride ( 64 CuCl 2 ) has been identified as a potential agent for PET imaging and radionuclide therapy. 78) In a study using 64 CuCl 2 relevant to radionuclide therapy, it was demonstrated that Cu metabolism is important for many cancers. Here, it is worth noting that compared with 64 Cu-labeled complexes, 64 CuCl 2 has simple radiochemistry without a radiolabeling process. The results prompted us to measure the biodistribution of 67 CuCl 2 in colorectal tumor-bearing mice. Colorectal cancer is a major cause of death in Japan. 79) 67 Cu was produced by irradiating an 68 ZnO (99.935% enriched in 68 Zn) sample with neutrons at TIARA-QST. 42) The chemical separation of 67 Cu from a neutron-irradiated 68 ZnO sample was performed by slightly modifying the previously reported method to separate Ga ions. 77) The radionuclide purity of 67 Cu was 99.8% at the time of injection. The specific activity [MBq/(µg Cu)] of 67 Cu was determined to be 4.5 MBq/(µg Cu) at EOI by the titration method. This value is much smaller than the typical specific activity of 64 Cu produced by the 64 Ni(p,n) 64 -Cu reaction in the range of 2.4-11 GBq/(µg Cu) quoted from a recent study on the biodistribution of 64 CuCl 2 in rats. 78) It is considered that the specific activity plays an important role in radiolabeling and the in vivo biodistribution of radioactive tracers. Hence, it is very interesting to study the role of 67 CuCl 2 with low specific activity in the biodistribution of 67 Cu ions in colorectal tumor-bearing mice. Animal procedures were carried out according to a protocol approved by the QST Institutional Animal Care and Use Committee. The 67 Cu solution after radiochemical purification was diluted with a physiological saline solution for injection into mice. When the tumors were palpable, the mice were intravenously injected with 35 or 50 kBq of 67 CuCl 2 dissolved in 100 µl saline via a tail vein. After the mice were sacrificed at 0.5, 1, 4, 8, 24, and 48 h postinjection (n F 4, four mice at a time), their blood and organ samples of interest (liver, kidney, intestine without content, spleen, pancreas, stomach, heart, lung, muscle, bone, brain, and tumor) were removed and weighed. 80) The radioactivities in the blood and organ samples were measured by using a well-type NaI(Tl) detector. The biodistribution of 67  bearing mice was determined, as shown in Fig. 20. Note that it is common in animal studies to express the biodistribution of radiotracers using the parameter %ID/g of tissue, defined as the radioactivity in a particular tissue at each time point as a percentage of the total radioactivity injected into the animal, which was further divided by the weight of each tissue. It is very interesting that a high uptake of 67 Cu in the tumor was found, which may indicate an important role of Cu metabolism in colorectal cancer. The accumulation of 67 Cu in the tumor was 7.0 ' 1.4%ID/g at 48 h, comparable to that of 64 Cu, 95%ID/g, in spite of the difference in the specific activities. A high uptake of 67 Cu was also observed in the organs, such as the liver and kidney. The 67 Cu uptake in the liver and kidney gradually decreased over time from 0.5 to 48 h. The biodistribution of 67 CuCl 2 determined by using very low-specific-activity 67 Cu is similar to the recent biodistribution of 64 CuCl 2 obtained by using high-specific-activity 64 Cu in malignant melanoma tumor-bearing mice. 78) The observed uptake of 67 Cu in these organs is considered to be due to copper metabolism being independent of the specific activity of 67 Cu.
In summary, 67 CuCl 2 was used for the first time to determine the biodistribution in colorectal tumorbearing mice. A high uptake of 67 Cu in the tumor was found, although the specific activity of 67 Cu was low owing to the neutron intensity currently available.
This result suggests that 67 CuCl 2 can be a potential radionuclide agent for cancer radiotherapy.
5. Deuteron accelerator and neutron source 5.1. Deuteron accelerator. Deuteron beams, provided mostly by linear accelerators, are used to produce high neutron fluxes by irradiating a light element, such as carbon, liquid Li, and Be, for example in the projects SPIRAL2, the Soreq Applied Research Accelerator Facility (SARAF) project in Israel, 81) and the International Fusion Materials Irradiation Facility (IFMIF) 82) et al., where a fixed neutron energy of 14 MeV is mostly needed. However, in order to produce a particular medical radioisotope by neutron-induced reaction using accelerator neutrons, a wide variety of the neutron energy is necessary, because a neutron-induced reaction cross section on samples depends on the neutron energy. Therefore, the neutron energy needs to be easily tuned to an energy suitable for the production of medical radioisotopes by changing the energy of the deuteron beams. In addition, we will use the same accelerator to produce not only accelerator neutrons, but also proton and deuteron beams, for generating a wide variety of medical radioisotopes via proton-and deuteron-induced reactions on a sample. In considering that accelerators have an active lifespan of over 30 years, and that interest in new medical radioisotopes will be continuously growing, the expected newly installed accelerators must have a capability for producing a wide variety of medical isotopes. We choose AVF cyclotrons with 50 MeV, 2 mA deuteron beams to meet the mentioned requirements. A fixed radiofrequency AVF cyclotron is robust in operation, compact in size, and relatively cheap compared to a linear accelerator. Such cyclotrons can be constructed by many cyclotron companies around the world. In fact, Sumitomo Heavy Industries, Ltd. has been constructing AVF cyclotrons, which can provide a 30 MeV, 1 mA H ! beam for Boron Neutron Capture Therapy (BNCT). 83) Here, it is worth mentioning that the beam intensities from cyclotrons are limited by an extraction device (deflector), and therefore negative deuteron D ! ions should be accelerated up to 50 MeV. The principal advantage of a D ! ions cyclotron is the ease and low loss in extraction by the stripping of D ! ions into positive deuteron (D D ) ions on a thin carbon foil with a thickness of about 500 µg/cm 2 . The D D ions can be extracted to a beam line through a residual magnetic field in the AVF cyclotron.

Accelerator neutron source.
So far, various types of accelerator-based neutron sources with high neutron fluxes that have kinetic energy above a few MeV have been developed for fundamental studies in nuclear physics and nuclear astrophysics, radiation-resistant materials irradiation testing for fusion reactors, boron neutron capture therapy, and slow neutron scattering etc. In these cases, accelerator neutrons are generated by the 7 Li(p,n) 7 Be, 3 H(d,n) 4 He, and 9 Be(p,n) 9 C reactions, and the spallation reaction by bombarding a liquid mercury target, a liquid bismuth-lead target with high-energy proton beams.
In medical radioisotope productions in reactors, thermal neutron fluxes of a factor # 10 14 n/(cm 2 s) have been used. Accelerator neutrons with a quasimonoenergy of 14 MeV with about 10 12 n/(cm 2 s) were used in the field of nuclear engineering at Fusion Neutronics Facility (FNS) of the Japan Atomic Energy Agency (JAEA), 51) and at a facility with a rotating target neutron source (RTNS-II) in U.S.A., 84) respectively. Because the 3 H target is radioactive, and neutrons from the 3 H(d,n) 4 He reaction at E d F 300 keV are emitted isotopically with respect to the deuteron beam direction, this leads to a disadvantage concerning the effective use of neutrons for RIs productions. Another intense neutron source based on the deuteron breakup reaction was proposed by P. Grand and A. N. Goland. 85) Note that Helmholz et al. first observed the breakup process, 86) and also found that an intense forward-directed beam of neutrons is emitted when a target with a low-atomic number, such as Li or Be, is bombarded with deuteron beams. They also designed a high-flux neutron generator system composed of a 35 MeV high-current deuteron linear accelerator and a molten Li target configuration. Based on the proposed neutron source, at the IFMIF, intense neutron fluxes of greater than 10 15 n/(cm 2 s) with the energy spectrum peaking at around 14 MeV are expected to be produced by bombarding a liquid lithium jet target with intense deuteron beams of about 35-40 MeV. At the SPIRAL2 facility, neutrons with a high flux of 10 15 n/s are planned to be produced by nat C(d,n) using 40 MeV 5 mA deuterons provided from a linear accelerator. At the SARAF facility, there is an ongoing project with the superconducting light/heavy-ion LINAC, with a potential of about 40 MV, capable of accelerating 5 mA deuterons up to 40 MeV. The neutrons are produced by irradiating deuterons on a liquid Li target.
We have also installed a mini-type rotating carbon target system used for producing accelerator neutrons provided by the nat C(d,n) reaction in collaboration with Sumitomo Heavy Industries, Ltd. A schematic view of the neutron target is shown in Fig. 21. The carbon target was shown to work successfully under a thermal power of 40 kW using the JAERI Electron Beam Irradiation System (JEBIS) at Japan Atomic Energy Agency (now QST), which can provide 20-100 keV electron beams with an output beam power of 400 kW. 87)

Conclusions and future prospects
We proposed an innovative method to produce a wide variety of medical radioisotopes. By overcoming challenges in the proposed method by fundamental  Tc suitable for formulating 99m Tc radiopharmaceuticals. In addition, we could validate the high capability of accelerator neutrons so far unexplored to produce a large amount of high-quality therapeutic radioisotopes conducting detailed studies of 64 Cu and 67 Cu. In order to secure a constant and reliable supply chain of 99 Mo for domestic use and to promote a theranostics approach in personalized nuclear medicine, we have presented a proposal for a prototype facility for the Generation of Radioisotopes with Accelerator Neutrons by Deuterons (GRAND). 46) The proposed system consists of an AVF cyclotron with a 50 MeV, 2 mA deuteron beam intensity and a rotating carbon target system to produce intense accelerator neutrons. In the cyclotron negative deuteron ions (D ! ) are accelerated up to 50 MeV, and by passing them through a stripper foil, D ! beams are converted to D D beams, which are extracted from the cyclotron to a beam-transport system for irradiating a carbon target to produce accelerator neutrons. The principal advantage of a negative deuteron-cyclotron is the ease and low loss in extraction by the stripping of negative deuteron ions into positive deuteron (D D ) ions on a thin carbon foil with a thickness of about 500 µg/cm 2 . The D D ions can be extracted to two beam lines through a residual magnetic field in the AVF with a possibility to irradiate two different targets simultaneously. The layout of the accelerator is shown in Fig. 22.
In April 2020, the two-year "Deuteron Accelerator for Theranostics mEdicine (DATE)" project at Tohoku University was started. In this project we plan to accelerate 25-40 MeV deuteron beams with an intensity of 100 µA by newly setting up a negative deuterium ion source and a stripper foil for the existing AVF cyclotron at CYRIC at Tohoku University. 34) The deuteron beam intensity will be about twenty-times stronger than the presently available D D beam intensity of about 5 µA, and about one-twentieth of the intensity that will be obtained at the GRAND project. At CYRIC, a negative hydrogen beam of 50 MeV with 22 µA was successfully obtained in 2003, which demonstrated the feasibility of high-current acceleration and extraction for proton beams with H ! acceleration. The DATE project will play an important role in the on-demand medical RIs production for promoting a theranostics approach in personalized nuclear medicine and also in detailed planning for a prototype facility, GRAND.