Cyclotron production of ²²⁵Ac from an electroplated ²²⁶Ra target

Abstract

Purpose

We demonstrate cyclotron production of high-quality ²²⁵Ac using an electroplated ²²⁶Ra target.

Methods

²²⁶Ra was extracted from legacy Ra sources using a chelating resin. Subsequent ion-exchange purification gave pure ²²⁶Ra with a certain amount of carrier Ba. The radium target was prepared by electroplating. We successfully deposited about 37 MBq of ²²⁶Ra on a target box. Maximum activation was achieved using 15.6 MeV protons on the target at 20 µA for 5 h. Two functional resins with various concentrations of nitric acid purified ²²⁵Ac and recovered ²²⁶Ra. Cooling the intermediate ²²⁵Ac for 2–3 weeks decayed the major byproduct of ²²⁶Ac and increased the radionuclidic purity of ²²⁵Ac. Repeating the same separation protocol provided high-quality ²²⁵Ac.

Results

We obtained ²²⁵Ac at a yield of about 2.4 MBq at the end of bombardment (EOB), and the subsequent initial purification gave 1.7 MBq of ²²⁵Ac with ²²⁶Ac/²²⁵Ac ratio of < 3% at 4 days from EOB. Additional cooling time coupled with the separation procedure (secondary purification) effectively increased the ²²⁵Ac (4n + 1 series) radionuclidic purity up to 99 + %. The recovered ²²⁵Ac had a similar identification to commercially available ²²⁵Ac originating from a ²²⁹Th/²²⁵Ac generator.

Conclusion

This procedure, which involves the ²²⁶Ra(p,2n)²²⁵Ac reaction and the appropriate purification, has the potential to be a major alternative pathway for ²²⁵Ac production because it can be performed in any facility with a compact cyclotron to address the increasing demand for ²²⁵Ac.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00259-021-05460-7.

Introduction

Targeted alpha therapy (TAT), which is a therapeutic regimen by radiopharmaceuticals labeled with alpha emitters, has received great interest due to its clinical impact such as ²²³RaCl₂ and ²²⁵Ac-PSMA-617 [1, 2]. Compared to conventional therapy by beta emitters, alpha particles exhibit a high linear energy transfer (LET) in a short range (40–100 µm/5–9 MeV, [3]), which may provide a remarkable cytotoxic effect in a limited area of the target. Hence, unwanted radiation doses to other healthy tissues and organs may be limited. Actinium-225 (²²⁵Ac, α = 100%, T_1/2 = 9.92 days) is a promising radionuclide applicable to TAT since it breeds multiple descendants via the net decays of 4α + 2β in a relatively short period (Fig. 1). This property enhances the therapeutic effects of ²²⁵Ac-labeled compounds [4].

Fig. 1.

Decay chains for the 4n, 4n + 1, and 4n + 2 series

The accessibility of various radioisotopes for diagnostic nuclear medicine is well established. Both accelerators and nuclear reactors are used to produce medical-grade isotopes. The former is typically used to produce short-lived isotopes, while the latter is used for large-scale, centralized production. A shortage of actinium-225 is anticipated [5] because interest has drastically increased but large-scale commercial production is still in the development phase. Currently, the most realistic path is the natural decay product by a ²²⁹Th/²²⁵Ac generator system [6, 7]. However, only a few institutes such as the Joint Research Centre (JRC, Karlsruhe, Germany), Oak Ridge National Laboratory (ORNL, TN, USA), and the Institute of Physics and Power Engineering (IPPE, Kaluga Oblast, Russia) have such capabilities and their estimated total annual capacity of actinium-225 is approximately 63 GBq (1.7 Ci) [8]. Consequently, alternative production pathways are highly desired.

Many studies have investigated increasing actinium-225 production to meet the anticipated demand. Practical options include (1) high-energy protons on ²³²Th (spallation channel, [9]), (2) moderate energy protons on ²²⁶Ra (nuclear transmutation channel, [10]), and (3) high-intensity gammas on ²²⁶Ra (photonuclear/Bremsstrahlung channel, [11–13]). Among these, option (1) holds promise because ²³²Th is not a fissile material according to nuclear regulations. However, the reaction requires a projectile with an extraordinarily high energy and intensity (e.g., 100 MeV or higher proton). Consequently, few facilities can practically produce ²²⁵Ac via the spallation route. On the other hand, options (2) and (3) use ²²⁶Ra as their target material. Option (2) may be advantageous for actinium-225 production in general facilities. However, this material is difficult to handle due to safety concerns (i.e., radon (²²²Rn) emanation and high-energy gamma emission from the descendants). For instance, as reported in [10], the reaction of ²²⁶Ra(p,2n)²²⁵Ac can be performed efficiently with a relatively small amount of ²²⁶Ra in a low-energy window provided by a compact medical cyclotron, Ep ≤ 20 MeV. In this study, we evaluate the production feasibility of ²²⁵Ac from a ²²⁶Ra target that includes (1) ²²⁶Ra recovery from legacy needles, (2) radium target preparation, (3) activation, (4) separation, and (5) recycling of ²²⁶Ra. This study should be useful not only from a production capability viewpoint but also from a quality control perspective for large-scale actinium-225 production.

Materials and methods

Materials

Hydrochloric acid (ultra-pure, 10 M) was purchased from Kanto Chemical (Tokyo, Japan). An ammonium acetate solution (10 M) was obtained from Nacalai Tesque (Kyoto, Japan). An ammonium solution (25%), nitric acid (70%), pure water, Dowex Monosphere 550A anion exchange resin (OH form, 590 ± 50 µm), and Dowex AG1-X8 anion exchange resin (Cl form, 100–200 mesh) were obtained from FUJIFILM Wako Chemicals (Tokyo, Japan). These reagents were used as received or diluted with the appropriate volume of pure water, as needed. Chelex-100 (Na form, 100–200 mesh) was purchased from Bio-Rad Laboratories (Tokyo, Japan). It was preconditioned as the ammonium form before use. Actinium-225 nitrate (37 MBq, 99.99% radionuclidic purity) was purchased from Oak Ridge National Laboratory and used as an authentic ²²⁵Ac source.

Methods

General

All procedures were performed in a ventilated glove box with a pressure of –50 Pa. No system for ²²²Rn handling was installed. A bag-in/out protocol with a polyethylene bag (thickness 100 µm) was employed when transferring samples across the glove box to avoid releasing ²²²Rn and other possible radioactive materials into the laboratory [Online Resource 1–3]. The maximum daily permission for handling ²²⁶Ra in our laboratory is 148 MBq (4 mCi).

Ra recovery from legacy needles

Radium needles (only information available is its size (Ø1.6 × 25 mm) and activity (1–2 mCi-²²⁶Ra/needle)) were sectioned into 5–6 pieces by an ordinary tube cutter (nipper type, for 1/16″ stainless tubes). The pieces, which were collected in a 50-mL glass bottle with a polypropylene screw cap (Duran Wheaton Kimble, Germany), were mixed with 3 mL of a Chelex-100 resin slurry and 7 mL of pure water. The tightly capped bottle was sonicated daily for a period of 1 week to 1 month.

Afterward, the Chelex-100 resin was filtered from the mixed materials by an empty cartridge (Bond Elut, 5 mL, Agilent Technologies, CA, USA), where the extracted ²²⁶Ra adsorbed on the resin. Then, 1 M HCl (5 mL) and subsequent pure water (10 mL) for rinsing were loaded into the cartridge to elute ²²⁶Ra from Chelex-100. The eluate was loaded into an anion exchange resin (16 mL, Monosphere) to remove chlorides. Then, the resin was washed with 10 mL pure water. The recovered solution was evaporated at 130 °C under a vacuum, yielding dried ²²⁶Ra in the hydroxide form.

Extending the treatment time improved the recovery efficiency of ²²⁶Ra. In our preliminary experiment, a maximum of 50% of the initial activity was recovered by the 1-week treatment. However, returning the mixture of needle pieces and Chelex-100 residue to the bottle again and repeating the same procedure for a month showed quantitative recovery of the initial activity. In the 1-month experiment, the slurry was adjusted to a pH ≥ 10 by adding a portion of conc. ammonium solution to ensure the Chelex-100 efficiently for absorbing ²²⁶Ra.

Target box design and target preparation

The ²²⁶Ra target was prepared by electrodeposition. Figure 2 shows the target box assembly. A Ti cylindrical cavity (#3 in Fig. 2) with a volume of about 3.5 mL was used for the electroplating reservoir and the dissolving vessel for the Ra target after activation. A Pt rod (Ø3 mm, anode for electroplating) was held by a polyimide screw with O-rings (#7). The bottom of the Ag cavity (#5) was assembled with both #3 and a polyimide electric insulator (#4). For chemical resistance, it should be noted that the Ra-depositing surface with a conical shape (#5) was fabricated with Au by hot isostatic pressing on the Ag body.

Fig. 2.

²²⁶Ra target assembly

Purified dried ²²⁶Ra, which ranged from 14.8 to 38.9 MBq (400–1050 µCi), was dissolved in 1 mL of 0.1 M HCl and 2 mL of 0.5 M ammonium acetate to prepare the electrolyte. The electrolyte was placed in the target box, and a constant current of 100 mA DC was applied in the pulse mode (5 Hz with a 0.1 s on–off cycle) for 3 h with a 15-mm gap between the cathode and the anode. After the electrodeposition process, the electrolyte was removed from the cavity by pipette work, and introduced in and removed from 2 mL of pure water twice to wash out the residual electrolyte in the target box. These rinsing solutions were collected as they may contain undeposited free ²²⁶Ra. Then, the deposition efficiency was evaluated by measuring the ²²⁶Ra activity. The cavity stood undisturbed overnight (> 15 h) to dry the Ra surface naturally in a ventilated glove box. Eventually, the Pt anode was withdrawn from the cavity and sealed with a thin Nb foil (50 µm). The cavity at the beam entrance was sealed with a 50-µm-thick Nb foil (#2 in Fig. 2) with #1.

Activation

Activations were carried out by 34 MeV H₂⁺ (ionized molecular hydrogen) provided by NIRS-AVF-930 cyclotron at a nominal intensity of 10 µA for 3–5 h. This condition increased the intensity of the lower energy particles accelerated by a relatively larger cyclotron to give 17 MeV protons at nearly 20 µA by splitting the kinetic H₂⁺ ion at the vacuum isolation window. The estimated proton energy on the target material by SRIM code [14] was 15.6 MeV after passing through the vacuum foil (Al, 100 µm), the He cooling layer (30 mm), and the target foil (Nb, 50 µm). To enrich the expected ²²⁵Ac yield, the on-target energy of 15.6 MeV was set between two energies showing the highest cross-sections for the ²²⁶Ra(p,2n)²²⁵Ac provided by the ALICE code (ca. 700 mb at 15 MeV) and the previous study (ca. 710 mb at 16.8 MeV) [10].

Separation of ²²⁵Ac from the target matrix

Figure 3 shows our newly developed separation procedure, which was implemented 3–4 days after the end of bombardment (EOB). The activated target was dissolved in 3 mL of 0.7 M HNO₃, and the solution was loaded slowly into a DGA cartridge (N,N,N′,N′-tetra-n-octyldiglycolamide, 1 mL, Eichrom Technologies, IL, USA). To increase the leftover recovery of Ac/Ra, another 3 mL of 0.7 M HNO₃ was introduced into the target cavity and the rinsing fraction was loaded into the same DGA cartridge. This step was repeated twice.

Fig. 3.

Separation diagram for ²²⁵Ac from the ²²⁶Ra matrix

The DGA cartridge was washed with 20 mL of 0.7 M HNO₃ to remove residual ²²⁶Ra. Then, 5 mM HNO₃ (10 mL) was loaded into the DGA to elute ²²⁵Ac, which was the fraction collected in an intermediate reservoir. Subsequently, the crude ²²⁵Ac fraction was loaded into a LN cartridge (di(2-ethylhexyl)orthophosphoric acid, 2 mL, Eichrom Technologies); the cartridge was washed with 10 mL of 50 mM HNO₃ to eliminate trace amounts of ²²⁶Ra and subsequently well purged. All the above waste fractions were collected as the Ra recovery fraction, which was recycled for the next use. Eventually, ²²⁵Ac was stripped by loading 0.7 M HNO₃ (10 mL) and collected into another intermediate reservoir.

The actinium-225 solution in this separation step contained ²²⁶Ac (β 83%, EC 17%, α 6 × 10^–3%; T_1/2 = 29.4 h) as a byproduct because it was unavoidably generated via the ²²⁶Ra(p,n)-channel in our activation condition. To increase the radionuclidic purity of ²²⁵Ac, the intermediate product was allowed to cool for 2–3 weeks, which is equivalent to 10 half-lives or more for ²²⁶Ac. After the cooling, the above separation protocol was repeated as the secondary purification.

Although the twice purified ²²⁵Ac was free from or had negligible ²²⁶Ac contamination in 0.7 M HNO₃ (10 mL), it was too acidic for further use. Thus, an anion exchange resin (AG1-X8, 100–200 mesh, Cl form) was employed to exchange the counter anion of ²²⁵Ac with chloride and remove HCl from the product by evaporation of the above sample (130 °C under vacuum). The final product, which had a chemical form of ²²⁵AcCl₃, was reconstituted in 100 µL of 0.1 M HCl for further use. The recovery yield of ²²⁵Ac was quantitative.

Recycling mode of Ra

All fractions possibly containing ²²⁶Ra were collected into a single vessel. The solution was adjusted to a pH ≥ 10 by adding a conc. ammonium solution and then loading into a column filled with the Chelex-100 resin (0.5 mL, NH₄ form) to concentrate ²²⁶Ra. After washing the column with 10 mL of pure water, ²²⁶Ra was stripped by passing 1 M HCl (5 mL), and the eluant was led to an anion exchange resin column (16 mL, Monosphere, OH form). The Ra fraction was desalted by an anion exchanger, and an additional 10 mL of pure water was loaded to remove the residual ²²⁶Ra. The collected ²²⁶Ra with a volume of about 15 mL was subsequently evaporated at 130 °C under a vacuum to yield purified ²²⁶Ra in the hydroxide form, which was ready for the next use as the electrolyte.

Sample analysis

Gamma spectroscopy

Samples were subjected to a HPGe detector coupled with a well-calibrated 4096-ch multi-channel analyzer (39 cm², EGC15-185-R, Eurisys (Mirion Technologies), Lingolsheim, France); RZMCA, Laboratory Equipment, Ibaraki, Japan). A measurement uncertainty of 9% was obtained as a positive square root of the sum of the following contributing factors in quadrature: counting statistics (5%), geometrical error (5%), and detector efficiency (6%). Table 1 lists the decay data for the radionuclides of interest [15]. The detection limit for these nuclides was 3.7 Bq (0.1 nCi) with an acquisition period of 60,000 s or longer, which was equivalent to 1.2 × 10^–3% of ²²⁵Ac activity in the most sensitive case.

Table 1.

Production results

Run	#1	#2	#3
Beam condition (Ep = 15.6 MeV)	20 µA × 3 h	20 µA × 5 h	20 µA × 5 h
Ra deposition
226Ra, initial electrolyte	14.5 MBq (391 µCi)	36.4 MBq (984 µCi)	38.8 MBq (1.05 mCi)
226Ra, deposited	13.5 MBq (366 µCi)	35.4 MBq (956 µCi)	37.5 MBq (1.01 mCi)
Deposition rate (%)	94	97	97
Nuclides of interest* in the initially purified sample (kBq, decay corrected to EOB)
225Ac (150 keV, 0.6%)	522	2.23 × 103	2.43 × 103
226Ac (230 keV, 26.9%)	111	451	488
224Ac (215 keV, 52.3%)	Not detected	Not detected	Not detected
226Ra (186 keV, 3.64%)	Not detected	Not detected	Not detected
214Pb (352 keV, 35.6%)	Not detected	Not detected	Not detected
214Bi (609 keV, 45.5%)	5.2	13.5	33.3
135La (481 keV, 1.52%)	84.5	333	344
140La (487 keV, 43.9%)	0.0571	0.165	0.231
212Bi (727 keV, 6.58%)	Trace	Trace	Trace
208Tl (2615 keV, 99.8%)	Trace	Trace	Trace

^*Nuclear data presented in parenthesis are used for quantification[15]

Alpha spectroscopy

An alpha spectrometer (Alpha Duo, Ametek Ortec, Oak Ridge, TN) equipped with an ion-implanted-silicon charged-particle detector (Ultra-AS 450 mm², Ametek Ortec) and 4096-ch pulse-height analyzer (Maestro-32, Ametek Ortec) was used to acquire the alpha spectra. The spectrometer was calibrated with a mixed source (¹⁴⁸ Ga, ²⁴¹Am, and ²⁴⁴Cm; Eckert & Zeigler, Valencia, CA). An aliquot of the sample (about 370–1 k Bq (10–50 nCi) of ²²⁵Ac) was dropped on an Al sheet, and dried. The prepared sample without a cover was loaded inside the chamber. These samples were analyzed with an acquisition time of 1200 s.

Results and discussion

Ra liberation from legacy needles

In most cases, ²²⁶Ra prepared in the radium needle had a form of RaSO₄. Although ionic compounds are typically water soluble, group II sulfates, including RaSO₄, are practically insoluble in water. Our samples showed a nearly zero recovery of free ²²⁶Ra²⁺ when the Ra matrix was suspended in water or 1 M HCl, providing additional evidence that Ra was in the sulfate form. However, when Chelex-100 was allowed to sit long term, a remarkable recovery occurred. Trace amounts of Ra²⁺ were gradually liberated from the sulfate as the chelation sites tightly held Ra²⁺.

In addition, the Ra matrix remained in the small sheath cavity, even though the legacy needle was cut into small pieces. Sonication seemed to crumble the solid Ra matrix, and free Ra was effectively released from the matrix. The recovery rate of ²²⁶Ra was 30–50% after a week but quantitative after a month.

Electrodeposition and activation results of ²²⁶Ra

The deposition yield of ²²⁶Ra was satisfactory. The difference of the ²²⁶Ra activity in the electrolyte between the initial and the post-deposition indicated that the yield ranged from 94 to 97% (Table 1). The deposited ²²⁶Ra layer, which contained some amount of carrier Ba, was practically insoluble upon washing after the deposition process. The washing fractions of pure water showed a very small activity of ²²⁶Ra (1–2% of the initial value).

Figure 4 shows the electrodeposition profiles of 1-mCi of ²²⁶Ra (ca. 4.5 µmol) with an unknown amount of carrier and 5-mg of Ba (ca. 36 µmol) as an increasing challenge/evaluation for Ra deposition using a Ba surrogate. Due to safety regulations for ²²⁶Ra handling and our limited Ra inventory, this study involved a small amount of ²²⁶Ra. Consequently, we performed a cold experiment with Ba instead of ²²⁶Ra to evaluate the electrodeposition performance based on the widely accepted chemical similarity between Ra and Ba. In the case of a small amount of Ra, 13.5–37.5 MBq (366–1010 µCi) in this study, a spot-like deposition profile appeared across the cathode surface (Fig. 4a), whereas a condensed layer covered from the top of the cone to the middle of the slope in a Ba rich condition (Fig. 4b).

Fig. 4.

Deposition profile of (a) approximately 37 MBq (1 mCi) of ²²⁶Ra on the target surface and (b) 5 mg of Ba. Both electrodepositions are performed under the same conditions (100 mA constant DC for 3 h)

These results can be explained by the principle of electrodeposition. The gap between the two electro-rods around the top of the cone was the shortest and formed the best condition for electroconductivity in this system. Consequently, effective deposition was expected around the top, and most of the solute (Ra or Ba) tended to be centralized on the surface. These results are encouraging evidence to obtain rich and high-density Ra depositions when there is sufficient Ra in the electrolyte.

We obtained ²²⁵Ac at a yield of 522–2430 kBq (14–66 µCi, decay corrected) from a ²²⁶Ra target of 13.5–37.5 MBq (366–1010 µCi) irradiated by 20 µA protons for 3–5 h (Table 1). A previous study employing a ²²⁶Ra(p,2n)-channel reported the production yield of ²²⁵Ac with various amounts of ²²⁶Ra [10]. Although it is difficult to directly compare the previous study and our data due to the different conditions (amount of ²²⁶Ra prepared, chemical form, target area, target thickness, beam energy, beam intensity, and irradiation time), the previously reported yield for ²²⁵Ac/²²⁶Ra (% activity) was 13–44% without considering the activation condition. By normalizing, this reported yield of ²²⁵Ac was converted ca. 0.10–0.36 nCi-²²⁵Ac/µg-²²⁶Ra/(µA h) (except for the result of 12.5 µg-Ra). Our results showed 0.63–0.65 nCi-²²⁵Ac/µg-²²⁶Ra/(µA h) by applying the same correction. Considering the above-mentioned differences, both production yields are within an acceptable range, suggesting that our production capacity is similar to that of the previous study. Hence, our data support the feasibility of ²²⁵Ac production from an electroplated elemental ²²⁶Ra target. In addition, the above production index demonstrates the good stability and repeatability of our target activation capability.

Separation

The initial separation of the ²²⁵Ac sample contained ²²⁶Ac and other radionuclidic impurities (Fig. 5a). Similar to ²²⁶Ra, ²²⁶Ac is a 4n + 2 series radionuclide, which generates many descendants during the cooling period (Fig. 1). Hence, repeated separation as a secondary purification removed the 4n + 2 impurities to yield high-quality ²²⁵Ac. Although ²²⁴Ac (EC 91%, α 9%, T_1/2 = 2.8 h) should be co-produced via the ²²⁶Ra(p,3n)-channel (E_TH = 13.6 MeV), the half-life of ²²⁴Ac was too short to be detected at the end of separation at 4 days from EOB. However, the major distributions in the washing fraction and leftovers of the separation materials were a couple of ²²⁴Ac descendants with favorable gamma emissions in the 4n series, ²¹²Bi (T_1/2 = 61 min, 727 keV, 6.7%) and ²⁰⁸Tl (T_1/2 = 3.1 min, 2615 keV, 99%). Moreover, trace amounts were detected in the purified actinium-225 sample, providing evidence for ²²⁴Ac generation. The presence of ²¹²Bi and ²⁰⁸Tl in the Ac fraction was acceptable because Bi was partially similar to Ac in our separation conditions. As a result, other Bi isotopes, ²¹⁴Bi (originating from ²²⁶Ra) and ²¹³Bi (a descendant of ²²⁵Ac), should also be found in the initial actinium-225 fraction. Orphan ²¹⁴Bi should decay upon additional cooling (Fig. 5b). On the other hand, ²¹²Pb (T_1/2 = 10.6 h, 239 keV, 44%), the parent nuclide for ²¹²Bi, was not detected in the purified ²²⁵Ac samples. All the 4n series-nuclides with the potential to be the parent for ²¹²Pb (²²⁴Ac–²¹⁶Po) had shorter half-lives than ²¹²Pb. The exception was ²²⁴Ra, which was removed along with ²²⁶Ra. Hence, only the 4n + 2 series was considered the byproduct in the separation process.

Fig. 5.

Gamma spectrum for various ²²⁵Ac samples: (a) ²²⁵Ac product (initial separation) after 4 days from the end of bombardment (EOB), (b) ²²⁵Ac product (secondary separation) after 20 days from EOB, and (c) commercially available authentic ²²⁵Ac (generator-made)

Other notable byproducts were ¹³⁵La (EC, T_1/2 = 19.5 h) and ¹⁴⁰La (β, T_1/2 = 1.68 days). The former presumably originated from a carrier of natural Ba in the legacy Ra needle via the ¹³⁵Ba(p,n)-channel. However, the half-life of ¹³⁵La is much shorter than that of ²²⁵Ac. Thus, an appropriate cooling time should gradually decrease the impact of ¹³⁵La on the ²²⁵Ac even though the legacy Ra was not chemically purified. On the other hand, since the heaviest stable isotope of Ba is ¹³⁸Ba, the atomic mass of ¹⁴⁰La was too rich to be generated by proton activation, suggesting that fission on ²²⁶Ra may occur in our activation condition. In addition, ¹⁴⁰Ba (β, T_1/2 = 12.6 days), which is a parent nuclide for ¹⁴⁰La, could also be generated as another fission product. Unfortunately, we were unable to directly confirm the presence of ¹⁴⁰Ba because most of the characteristic gamma lines for ¹⁴⁰Ba were close to those for ²¹⁴Bi (RaC) and the chemical similarity between Ba and Ra. However, the initial separation of the actinium-225 fraction would practically eliminate ¹⁴⁰Ba along with ²²⁶Ra due to chemical similarity. Indeed, orphan ¹⁴⁰La in the actinium-225 fraction showed an acceptable half-life of 1.67 ± 0.10 days and decayed to a non-detectable level on the gamma spectrum by cooling for 2–3 weeks. This finding suggests that the carrier Ba in Ra needles does not affect the quality of ²²⁵Ac, and Ra purification from carrier Ba does not provide a practical advantage. A discussion on the counter fragments is available in [Online Resource 4].

For example, we cooled the samples for 19–20 days after EOB or 2 weeks from the end of separation. The spectra of the cooled samples were similar to that of authentic ²²⁵Ac originating from a ²²⁹Th/²²⁵Ac generator (Fig. 5b and c). The alpha spectrum of our ²²⁵Ac product also showed the same profile as the reference (Fig. 6). Notably, neither ²²⁶Ra (Eα = 4.78 MeV, 94%) nor ²¹⁰Po (Eα = 5.30 MeV, 100%) was detected. Hence, the double separation with an appropriate cooling period gave pure ²²⁵Ac with a quality comparable to generator-made ²²⁵Ac.

Fig. 6.

Alpha spectrum for the ²²⁵Ac product. Aliquot of 0.37–1.85-kBq ²²⁵Ac dried on an Al disk measured without a covering for (a) purified ²²⁵Ac product after 19 days from the end of bombardment (EOB) and (b) commercially available authentic ²²⁵Ac

Recovery of ²²⁶Ra for recycling

We developed a closed circuit for Ra recycling to minimize the loss of the ²²⁶Ra inventory (Figs. 3 and 7). This process effectively reduced long-lived radioactive wastes. After single-runs of this circuit with a 37-MBq (1-mCi) ²²⁶Ra, we evaluated the ²²⁶Ra leftover in each measurable material (i.e., the Ra recovery fraction), separation material (cartridge), and reservoir. The Ra recovery fraction contained 90–98% ²²⁶Ra, and other materials were negligible. For example, < 37–74 kBq (1–2 µCi) of ²²⁶Ra activity was found in the respective materials, depending on the volume of the residual liquid presented in the small voids. In addition, any deposition/leftover on the target box could not be estimated due to its high radioactivity. This can explain the ~ 10% discrepancy in the activity distribution. Since the target box was used repeatedly, the practical loss of ²²⁶Ra should be negligible.

Fig. 7.

Diagram of ²²⁶Ra recycling

Radionuclidic purity—impact of ²²⁶Ac in ²²⁵Ac

Actinium-226 disintegrates by pathways of β decay (83%), EC (17%), and trace α decay (6 × 10^–3), which breed ²²⁶Th (α, T_1/2 = 30.6 min), ²²⁶Ra, and ²²²Fr (β, T_1/2 = 14.2 min), respectively. Among these radioisotopes, ²²⁶Ra, which has the longest half-life, may be concern when preparing ²²⁵Ac-labeled injections. Although radiation safety related to both ²²⁶Ac and ²²⁶Ra is beyond the scope of this study, the amount of ²²⁶Ra generation as a decay product of ²²⁶Ac should be considered in discussions on clinical applications or the appropriate cooling time to address the acceptable quality on the actinium-225 product.

Actinium-226 with activities of 3.2 MBq (86 µCi) and 0.24 MBq (6.4 µCi) may generate 1.13 Bq and 0.084 Bq of ²²⁶Ra, respectively. These values are equivalent to ²²⁶Ra in the human body estimated (31 pg, where 27 pg is accumulated in the skeletal system) as well as those from dietary intake (2.3 pg/day) [16]. According to the reference [2], a maximum realistic single dose of actinium-225 injection is around 10 MBq (100 kBq/kg). Therefore, a ²²⁶Ac/²²⁵Ac sample in a 10-MBq actinium-225 injection may generate ²²⁶Ra equivalent to the above-mentioned reference.

While the targeted alpha therapy field has yet to establish an acceptable limit for ²²⁷Ac (β 98.6%, α 1.38%; T_1/2 = 21.8 y) [17], which is a major byproduct in the spallation pathway from ²³²Th target, ²²⁷Ac should be a good reference for rational considerations for long-life alpha-emitting byproducts. A previous study reported that ²²⁷Ac was equivalent to 0.7% of ²²⁵Ac at the time of injection (radioactivity-based estimation) [18]. One study reported that the ²²⁷Ac contribution to the radiation dose delivered by an actinium-225 injection was negligible [19]. However, the presence of ²²⁷Ac may pose a waste disposal issue. A recent study demonstrated the production feasibility of ²²⁵Ac without ²²⁷Ac (²²⁷Ac/²²⁵Ac =  < 7.5 × 10^–5%) using ²²⁵Ra (β 100%; T_1/2 = 14.9 d), which is another spallation product from ²³²Th, as the parent nuclide [17]. This report referenced the exemption activity recommended by the International Atomic Energy Agency (IAEA) [20] for a rational discussion about the long-life radionuclide.

Although our production yields were smaller than the assumed single dose of 10-MBq ²²⁵Ac, the activity ratio of ²²⁶Ac/²²⁵Ac in our samples ranged 1.4–2.3% at the end of the initial separation (4 days from EOB). The ratio was comparable to the potential amount for ²²⁶Ra delivery via dietary intake (²²⁶Ac/²²⁵Ac = 2.4%). The ²²⁶Ac content of around 1–2% (100–200-kBq ²²⁶Ac/10-MBq ²²⁵Ac) was higher than the exemption activity for ²²⁶Ac (100 kBq, [20]) but additional cooling for several days would gradually decrease the ²²⁶Ac activity below the exemption. In addition, the potential ²²⁶Ra amount in our purified samples at any time was much lower than the exemption activity (10 kBq, [20]), suggesting that radiation risks caused by ²²⁶Ac would be negligible or exceedingly small, if the ²²⁶Ac/²²⁵Ac ratio is close to the range of our results.

As shown in Fig. 5b, secondary separation effectively increased the radionuclidic purity of ²²⁵Ac, which reached > 99% within 2–3 weeks. In the future, we plan to evaluate the biological impact of ²²⁶Ac since the physical decay loss of ²²⁵Ac during the cooling period is a critical issue in the actinium-225 industry.

Conclusion

Actinium-225 purified from an electro-deposited ²²⁶Ra target with two separation columns showed an acceptable quality without byproducts. The characteristics of the purified ²²⁵Ac were similar to those of commercially available ²²⁵Ac originating from a generator system. The production results showed a linear increase in the ²²⁵Ac yield by increasing ²²⁶Ra prepared.

Consequently, increasing the ²²⁶Ra, beam intensity, or irradiation period can achieve the clinical requirement of ²²⁵Ac yield, demonstrating that the proposed production method may be a viable alternative pathway to address the increasing demand for actinium-225.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contribution

Conceptualization [Kotaro Nagatsu]; data curation [Kotaro Nagatsu]; formal analysis [Kotaro Nagatsu]; funding acquisition [Kotaro Nagatsu, Tatsuya Higashi, Ming-Rong Zhang]; investigation [all authors]; methodology [Kotaro Nagatsu, Hisashi Suzuki, Masami Fukada, Taku Ito, Jun Ichinose, Yoshio Honda, Katsuyuki Minegishi]; project administration [Kotaro Nagatsu, Taku Ito, Tatsuya Higashi, Ming-Rong Zhang]; resources [all authors]; software [N/A]; supervision [Kotaro Nagatsu]; validation [all authors]; visualization [Kotaro Nagatsu]; writing – original draft [Kotaro Nagatsu]; writing – review and editing [all authors].

Funding

This work was supported by JSPS KAKENHI [Grant Numbers JP26461814 (K.N.), JP17K10384 (K.N.), and JP20K08096 (K.N.)] and by AMED [Grant Number JP17pc0101014 (T.H.)].

Declarations

Ethics approval

Ethical approval is not required because no biological materials are used.

Consent to participate

Consent is not required as the experiment did not involve people.

Conflict of interest

The authors declare no competing interests.