Targeted alpha (α) therapy (TAT) is becoming a promising way to treat incurable cancers1. An α-emitter conjugated with a suitable carrier can transfer lethal energy to cancer cells precisely even when the tumor tissue is tiny. TAT minimizes damages in normal tissues due to high linear energy transfer (60–230 keV/μm) and short range (50–90 μm) of α-particles in human body. Demand for α-emitters has been increasing since the US FDA approved 223Ra chloride as a radiopharmaceutical. For cancer treatments, both 225Ac and 227Th are regarded as best α-emitters due to an adequate half-life and emission of multiple α-particles by daughter nuclides. After an outstanding efficacy was reported for 225Ac in treating prostate cancers, a lot of researches on TAT with 225Ac are underway globally. Meanwhile, 227Th is also getting lots of attention due to its chemical affinity to be easily bonded with various antibodies. In particular, 227Th decays to 223Ra, which is an officially approved α-emitter for clinical use, and it is rather free from regulatory issues other than its daughter2,3,4. Radiopharmaceuticals labelled with 227Th are being tested for various cancer treatments as well2,5.

Despite of the high potential of 225Ac and 227Th, lack of their supply is impeding progress in radiopharmaceutical for TAT. Even for 225Ac, the annual supply is only about one thirtieth of its demand that reached about 1850 GBq in 20196. The only commercial way to produce 225Ac is a milking from 229Th cow which is a decay product of the legacy 233U stockpile6,7. It is known that the stockpile was mainly produced in the Th-fuelled nuclear reactors in the 1960s in US and Russia8,9. Consequently, the 225Ac supply is rather monopolized and affordable ways are strongly required for progress in the 225Ac-based TAT.

Researches are currently underway to find out alternative methods to produce 225Ac. For example, the Tri-lab project team is looking into ways to directly extract 225Ac from a proton-irradiated thorium target, and produced 81.4 GBq of 225Ac in a 10-day irradiation of 200 MeV proton beam at 165 μA10,11. Regarding this method, one concern is ~ 0.3% yield of 227Ac contaminating 225Ac in addition to the relatively high proton energy required. It is known that 227Ac may cause severe level of radiation dose for the rest of patient’s lives if it is injected with 225Ac12. Meanwhile, utilizing 226Ra(p,2n)225Ac reaction is regarded as the most feasible approach due to a low proton energy. Institute for Transuranium Elements (ITU) produced ~ 485 MBq of 225Ac by irradiating a RaCl2 target with a 28 MeV beam for 45.3 h at 50 μA beam current7,13. However, it is known that procurement and treatment of the highly radioactive Ra target are challenging and costly. Though other methods including irradiation in nuclear reactors have been investigated to utilize 232Th(n,γ)233U and 226Ra(n,2n)225Ra reactions, low yield of 225Ac and 227Ac impurity catches up with their commercialization7.

The situation is even worse in the case of 227Th production. The natural 227Ac yield from 235U is very small due to long half-life of 235U and 227Th briefly exists as a daughter nuclide of 227Ac. Currently, there are no other options but to produce 227Ac from the 226Ra(n,γ)227Ra reaction2,14, which means that the aforementioned Ra target issue remains unresolved and a costly neutron source is required.

In response to the high demand of the ‘rarest drugs’ 225Ac and 227Th, we propose an innovative method to semi-permanently produce the α-emitters with an electron accelerator and natural thorium (232Th) target. The Monte Carlo N-Particle transport (MCNP) version 6.2 code is used to calculate X-ray or photon generation and heat deposition in the Th target15. The photo-transmutations of 232Th into 225Ac or 227Th are estimated by using the TALYS-generated Evaluated Nuclear Data Libraries (TENDL) cross-sections for various (γ,xn) photonuclear reactions16.


Photonuclear reactions in thorium target

Figure 1 shows schematic diagram of the proposed Th target system for photo-production of both 225Ac and 227Th. A metallic cylindrical Th target is directly bombarded by an electron beam to generate Bremsstrahlung photons and trigger photonuclear reactions. The target should rotate to manage locally-deposited heat during beam irradiation and it is supposed to be equipped with a cooling and radiation protection system as in the TRIUMF Isotope Separator and Accelerator (ISAC) facility17. Taking into account target cooling, the electron beam power is set to 500 kW and a relatively large target is considered (radius = 10 cm and thickness = 6 cm) to minimize the photon leakage from the target. To achieve appropriate target cooling, we adopt a beam diameter of 40 mm and a rotational speed of 120 RPM (rotation per minute). Detailed discussion on the heat deposition and removal is provided in Supplementary Information Sect. A.

Figure 1
figure 1

Schematic diagram of thorium target system.

Figure 2 describes major photonuclear reactions and transmutation chains of 13 nuclides in the Th target. Two most significant pathways to generate 225Ac and 227Th are highlighted with bold lines respectively: 232Th(γ,3n)229Th → 225Ra → 225Ac and 232Th(γ,n)231Th → 231 Pa → 227Ac → 227Th. Isotopic yields involving multiple photonuclear reactions with other thorium isotopes are actually very small. Other actinium isotopes can also be generated from (γ,p) and (γ,np) photonuclear reactions. However, they quickly decay to thorium isotopes except for 229Th(γ,np)227Ac reaction, which is insignificant due to very small (γ,np) cross-section. We also neglect the photo-fission of 232Th, in spite of noticeable probability, since both fission products and neutrons hardly affect yield of 225Ac and 227Th. Meanwhile, the impact of photo-fissions on heat deposition turns out to be noticeable and it is discussed in Supplementary Information Sect. A. In the Th target, many fast neutrons are also generated by the photonuclear reactions including the photo-fissions. When the Th target is bombarded with 500 kW electron beam (70 MeV energy and 7.14 mA current), the neutron production is roughly estimated to be ~ 1.94E + 15 #/s. However, it is found that most of the fast neutrons (> 97%) simply escape the target and cannot contribute to the production of 225Ac and 227Th.

Figure 2
figure 2

Photonulcear reactions and decays in Th target during electron beam irradiation.

Measured cross-sections of the photonuclear reactions of the Th and Pa isotopes are unknown except for (γ,n) and (γ,2n) reactions of 232Th18. In order to calculate the photonuclear transmutation rate, the TENDL nuclear data are utilized. Figure 3 shows experimental (γ,xn) and photo-fission cross-sections of 232Th together with the TENDL data for comparison. Except for overrated TENDL data of (γ,n) cross-section at a high energy tail, most of the experimental data for (γ,n) and (γ,2n) reactions are within one standard deviation of the TENDL data. While nuclear reaction modelling codes including TALYS generally provide rather uncertain parameters for photonuclear reactions, it is known that the parameter of giant dipole resonance (GDR) reactions such as (γ,xn) of heavy nuclides is slowly varying with the atomic mass number and the deviation from one nuclide to another is rather small19,20,21. Therefore, a GDR cross-section of a nuclide can be well predicted based on the experimental data of neighboring nuclei. Based on the measured cross-sections and general credibility of the TENDL library as shown in Fig. 3, actual cross-sections of 232Th(γ,3n)229Th and other (γ,xn) reactions with similar mass number such as 231Pa are expected to be similar to the TENDL data. Impacts of TENDL data uncertainty on the isotopic yields are given in Supplementary Information Sect. E.

Figure 3
figure 3

Comparison between TENDL and experimental data for (γ,xn) cross-sections of 232Th.

As indicated in Fig. 3, high energy (> 6 MeV) photons are required for the photonuclear (γ,xn) reactions of 232Th and the electron energy should exceed about 30 MeV. The Bremsstrahlung photon source in the Th target is generated by using the well-validated MCNP code. For a given photon source, number density of ‘A’ nuclide (NA) is estimated by solving the following balance equations.

$$ \frac{{dN^{A} }}{dt} = - \lambda_{Nat}^{A} N^{A} - \lambda_{\gamma }^{A} N^{A} + \lambda_{Nat}^{Other} N^{Other} + \lambda_{{(\gamma ,{\text{x}}n)}}^{Other} N^{Other} , \, $$
$$ \lambda_{{(\gamma ,{\text{x}}n)}}^{A} = \left( {\int {\sigma_{{(\gamma ,{\text{x}}n)}}^{A} \frac{{dN_{\gamma } }}{dE}dE} } \right), \, \lambda_{\gamma }^{A} = \sum\limits_{{{\text{x}} = 1}}^{3} {\lambda_{{(\gamma ,{\text{x}}n)}}^{A} , \, } $$

where λNat is natural decay constant, λ(γ,xn) is an effective decay constant for a (γ,xn) reaction, λγ is summation of the effective decay constants for all (γ,xn) reactions, and dNγ/dE is the photon spectral density obtained by MCNP 6.2. A space-dependent photon spectral density should be used for accurate evaluation of the isotopic yields. The last two terms in Eq. (1) denote production of ‘A’ nuclide due to decay or transmutation of other nuclides. Detailed coupled equations for the 13 nuclides in Fig. 2 are described in Supplementary Information Sect. B. Spatial distribution of the photon source in Eq. (2) is pre-evaluated with MCNP and it is assumed to be constant since transmutation rate of 232Th is extremely low in the rotating Th target.

Procedures to retrieve pure 227Th and 225Ac

To produce pure 225Ac and 227Th from the irradiated target, their parent nuclides, 229Th and 227Ac, should be separated in order to avoid the 227Ac contamination. There have been a few experimental validations for extraction of 225Ac from 233U or 229Th stockpiles22,23. For the same purpose, we propose a simple strategy and chemical processings for periodic milking of the α-emitters, as shown in Fig. 4.

Figure 4
figure 4

Procedure for semi-permanent milking of 225Ac and 227Th.

Purpose of the cooling phase is to let 231Th (progenitor of 227Ac) completely decay out. After irradiation, the target should be cooled for a certain time to minimize the 227Ac impurity for pure 225Ac extraction. Once 231Th disappears by its natural decay, all remaining Th isotopes are chemically separated from the others, then 225Ac is purely produced by decay of 229Th and this chemical extraction of 225Ac can be near permanent due to long half-life of 229Th. It should be noted that 226Ra can be also produced due to decay of 230Th in the separated Th stream and it can be used as the target material for other accelerated-based 225Ac production methods.

Meanwhile, pure 227Th can be chemically extracted from the separated irradiation byproducts, which are basically Ra, Ac, and Pa isotopes, i.e., 231Pa slowly decays to 227Ac decaying again to 227Th with a half-life of ~ 22 years. If 227Th is directly extracted from the separated byproducts without any additional chemical partitionings, the 227Th can be blended with 230Th generated by a natural decay of 230Pa, which may cause an unwanted hazard induced by its daughter nuclides. For the purity of 227Th, we chemically partition the byproducts into three parts: Ra, Ac, and Pa isotopes (blue lines in the non-Th products in Fig. 4). Then, 227Ac is extracted from the separated Pa and moved to the actinium part where pure 227Th can be purely extracted periodically. It is important to note that 227Th production is quite semi-permanent (over 50 years) since half-life of 227Ac is about 22 years. All chemical procedures in Fig. 4 are very analogous with ones used in the current 225Ac extraction from natural thorium target bombarded by proton beam24,25, and they are considered to be readily available. It is also obvious that the 225Ac and 227Th yields should increase if the amount of 229Th and 231 Pa in the target is increased with a prolonged beam irradiation and/or higher electron current.


Parameter optimization for electron beam

A 500 kW electron beam is considered to evaluate the feasibility and performance of the new photo-production of 225Ac and 227Th. In order to find optimal parameters for the 500 kW electron beam, isotopic yield of 229Th and 231Pa (parent nuclei of 225Ac and 227Th) is evaluated for different beam diameters and electron energies. For accurate evaluation of the photon spatial distribution, the target is subdivided into small zones in both radial and axial directions, which is described in detail in Supplementary Information Sect. C.

For the sensitivity analysis, it is assumed that a 70 MeV beam bombards a fixed target with a 5 cm radius and a 6 cm height. The target size is large enough to minimize the leakage of Bremsstrahlung photons from the target. Figure 5 shows the spatial distribution of isotopic yields with different electron beam sizes (yellow area) after a one-year beam irradiation and cooling of a month. The spatial isotopic yield is evaluated with a different spatial discretization of the target depending on the beam size, as indicated by the dotted lines in Fig. 5. One notes that total yield of 229Th drops by ~ 22% when beam diameter is increased from 15 to 40 mm, whereas the 40 mm beam enhances 231Pa production slightly. The clearly lower 229Th yield with the 40 mm beam is because contribution of the 231Pa(γ,2n)229Pa reaction can be noticeably lowered due to lower concentration of 231Pa with a larger beam diameter. Meanwhile, the slightly higher 231Pa production for the 40 mm beam is simply due to the slower depletion rate of 232Th in a bigger irradiated region. It is clear that a smaller beam size is favorable in terms of the 229Th yield if the integrity of the target is guaranteed during irradiation.

Figure 5
figure 5

Isotopic yield (MBq) in a fixed target with different beam diameters.

In the actual rotating Th target as shown in Fig. 1, the effective beam size or irradiated area is significantly increased compared with a fixed target. In the Th target with a 120 RPM speed, it is assumed that the spatial photon spectral density is uniform in the azimuthal direction and they are evaluated with a detailed discretization in the radial direction. The isotopic yields of 229Th and 231 Pa are given in Fig. 6 when the target radius is 10 cm and beam diameter is 40 mm. It is noted that total yield of 229Th is slightly lower than that of the fixed target due to further dispersion of the electron beam, and production of 231 Pa is slightly increased. The isotopic yields in the rotating target are compared with ones in the fixed target in Table 1.

Figure 6
figure 6

Isotopic yield (MBq) in the rotating target with 40 mm beam diameter.

Table 1 Isotopic yield (GBq) with different beam irradiation conditions.

Figure 7a shows MCNP-evaluated zone-wise photon spectral densities for the rotational target in Fig. 6. One clearly notes that the photon flux is highest in the zone A and decreases more quickly in the radial direction due to a localized electron distribution. The highly position-dependent photon density results in the space-dependent yield distribution in Fig. 6. The number of electrons in the MCNP simulation is 8,000,000, leading to accurate photon spectral density in the whole target region. It is mentioned that impact of uncertainty of the photon flux on the α-emitter yield is less than 0.6%.

Figure 7
figure 7

Zone-wise photon spectral densities for 70 MeV electron beam and isotopic yield of 229Th and 231 Pa as a function of electron energy.

Figure 7b shows the isotopic yield of 229Th (source of 225Ac) and 231Pa (source of 227Th) in the whole target as a function of the electron energy after a one-year beam irradiation and cooling of a month. The yield for both isotopes converges to an asymptotic value with electron energy since the photonuclear cross-sections of 232Th are maximized below 30 MeV as indicated in Fig. 3. The optimal energy is about 70 MeV for the 231Pa, while it is slightly above 90 MeV for 229Th. This is because 229Th production is mainly dominated by the (γ,3n) reaction requiring higher electron energy than in the (γ,n)-dominating 231Pa production. Based on the results, 70 MeV is considered to be near optimal to produce the two isotopes simultaneously as incremental yield of 229Th is marginal above 70 MeV.

Periodic yield of 225Ac and 227Th

Figure 8 shows time-dependent isotope-wise number density in the rotating target and activity of important nuclides in the periodic extractions. The 227Ac can be less than 10–8% in the separated Th stream after 30-day cooling as shown in Fig. 8a. Nuclides inventories in the separated Th and non-Th products for periodic extractions of 225Ac and 227Th are shown in Fig. 8b,c. Long-term extraction of both α-emitters can be realized due to a sufficient amount of their parent nuclei and it is clear that extraction of pure 225Ac and 227Th is possible in each separated stream. Optimal extraction or milking period for 225Ac and 227Th is determined so that accumulated yield should be maximized, as shown in Fig. 8d.

Figure 8
figure 8

Isotopic evolution in Th target, periodic extraction and accumulated yield of 225Ac and 227Th.

When 225Ac is extracted for 50 years after a one-year beam irradiation, its accumulated yield increases until 31-day cooling period and gradually dwindles. Meanwhile, there is no local maximum point of the accumulated 227Th yield with different extraction periods due to a large difference of half-life between 227Th and its parent nuclide, and basically the shorter the extraction, the better in terms of the accumulated yield. In this work, we regard 1 day for 227Th and 31 days for 225Ac as optimum extraction period.

Figure 9 shows the yield of 225Ac and 227Th per an extraction for three irradiation periods with a 70 MeV beam over a 50-year operation. One observes in Fig. 8 that 225Ac yield is almost constant, while the 227Th yield increases over 30 years and then slowly decreases due to the longer half-life of 227Ac than that of 225Ra decaying to 225Ac, which is the root cause of the low 227Th yield right after irradiation. If a daily 227Th yield is too small, a longer extraction period can be adopted in the early operational period (See Supplementary Information Sect. D).

Figure 9
figure 9

Yield of 225Ac and 227Th per extraction with different beam irradiation times.


Table 2 compares the proposed method with other approaches using a 226Ra target to produce 225Ac and 227Th. Performances of the new method are based on the mean TENDL data. Their variations due to uncertain TENDL data are discussed in Supplementary Information Sect. E . Regarding the existing methods, an optimistic 225Ac yield is also evaluated by improving beam and target parameters13,26,27. For an optimistic 227Th yield of the reactor-based approach14, an optimal irradiation time and a relatively large target are considered (see Supplementary Information Sect. F).

Table 2 Comparison of 225Ac and 227Th production methods.

The new method requires only one ‘long’ irradiation and 225Ac can be extracted semi-permanently. However, a ‘short’ irradiation is always necessary for each production of 225Ac in the others. In order to compare a yearly yield, it is assumed, based on engineering judgements, that there should be at least 12-h preparation time between irradiations in the conventional methods. The yearly 227Th yield is averaged over a 50-year operation since it is time-dependent in both new and reactor-based approaches.

The optimistic reactor-based 227Th yield is much higher than that of the new method due to the very favorable target and irradiation conditions. Unfortunately, the nuclear reactor is extremely costly and is not available in many countries. Although the yearly 225Ac yield of the new method is lower than the optimistic expectation of the others, the following advantages should be recalled: (1) pure 225Ac can be produced in the new photo-production, while there should be some level of 227Ac contamination in methods utilizing a Th target bombarded by a proton beam, (2) a semi-permanent source is produced after a single irradiation in a compact electron accelerator, (3) both 225Ac and 227Th are simultaneously produced. It is also noteworthy that both yields are rather proportional to the irradiation time since contribution of multiple photonuclear reactions to the yields is quite marginal in the rotational Th target.


The newly proposed 225Ac and 227Th photo-production method is deemed to be viable and promising in several aspects. First of all, it can produce pure 225Ac and 227Th by irradiating a ‘cheap’ natural Th target with a 50–70 MeV electron beam and then using the conventional chemical procedures. The electron energy needs to be over 50 MeV and 70 MeV turns out to be optimal for the application. To maximize the 225Ac and 227Th yield, a rotational Th target is required and it is shown that yearly yield of 225Ac and 227Th can be ~ 8.5 GBq and 49 GBq, respectively, for a 500 kW beam. The required electron accelerator is easily available and the new method can get all counties to benefit from TAT. More importantly 225Ac production is almost permanent and 227Th can be produced much over 50 years after a beam irradiation. Yearly 227Th yield is about four times higher than that of 225Ac, and the new method is particularly useful for production of emerging 227Th. We believe that this work will help increase global supply of the two precious isotopes and would invariably help advance TAT-related researches and developments.

Taking into account the photonuclear cross-sections, the expected 227Th yield seems to be rather reliable while the 225Ac yield is more uncertain due to absence of measurement data for the governing (γ,3n) cross-section of 232Th. Therefore, the 232Th(γ,3n)229Th cross-section should be experimentally evaluated for more concrete conclusions.