Introduction

Radioisotopes are widely used for the diagnosis and therapy of different diseases owing to their nuclear-physical properties [1, 2]. Diagnostic radioisotopes can provide functional and metabolic information for the early treatment of diseased regions that have not yet undergone morphological or structural changes. Currently, positron emission tomography (PET) [3] and single-photon emission computed tomography (SPECT) [4] are the two major diagnostic techniques. Short-lived β⁺-emitting radioisotopes are often used for PET. Typical examples include ¹¹C, ¹³N, ¹⁵O, and ¹⁸F, which have half-lives of 20, 10, 2, and 110 min, respectively. For SPECT, γ-ray emitting radioisotopes are frequently used, of which ^99mTc with a half-life of 6 h [5] is the most common radioisotope tracer. Therapeutic radioisotopes can be combined with targeted drugs to achieve the precise removal of small diseased regions without excessive doses to normal tissues. Therefore, radioisotopes that emit low-range highly ionizing radiation are of significant interest. The β^--particle emitting radioisotopes (e.g. ³²P, ⁸⁹Sr, ⁹⁰Y, ¹³¹I, ¹⁷⁷Lu, and ¹⁸⁸Re), Auger electron cascades (e.g. ¹⁰³Pd and ¹²⁵I) and α-particle emitting radioisotopes (e.g. ²¹¹At, ²¹²Bi, ²²³Ra, ²²⁵Ac, and ²²⁷Th), with a high linear energy transfer in tissue, are suitable for therapy.

More than 40 million nuclear medicine procedures are performed annually, and the demand for radioisotopes is increasing by 5% annually [6]. Currently, the production of medical radioisotopes relies primarily on nuclear reactors and cyclotrons. Thermal neutron-induced fission produces neutron-rich radioisotopes in nuclear reactors. The advantage of this method is the possibility of producing high levels of total and specific activities. However, the production of desired radioisotopes is accompanied by a considerable amount of long-lived radioactive waste, which raises numerous safety and security concerns. In addition, many nuclear reactors used for medical radioisotope production are more than 50 years old and may be shut down in the near future [7]. For example, a vast majority of reactors producing ⁹⁹Mo are expected to shut down by 2030. Compared with reactors, cyclotrons typically produce neutron-deficient radioisotopes by charged particle reactions accompanied by less radioactive waste. In addition, they have the advantage of being more compact, allowing their placement near hospitals, and making them a useful tool for producing radioisotopes with short half-lives that range from several minutes to hours.

Another promising method for radioisotope production involves photonuclear reactions, mainly including γ, n, (γ, p), and (γ,γ’). This is considered an alternative to radioisotope production in reactors and cyclotrons or the only method of production for certain radioisotopes. γ, n reactions can produce β⁺-emitting radioisotopes for PET imaging, such as ¹¹C, ¹³N, ¹⁵O, and ¹⁸F. These radioisotopes are produced by cyclotron-based (p,n) or (p, γ) reactions. (γ, p) reactions are suitable for producing β^--emitting radioisotopes that are currently mostly produced in reactors. High-intensity γ beams are required to obtain adequate yield using this method. With the development of laser Compton-scattering (LCS) gamma sources, the production of radioisotopes via photonuclear reactions has attracted considerable attention [8-17].

In contrast to conventional bremsstrahlung gamma sources based on electron linear accelerators, LCS gamma sources have the advantages of a high photon flux and excellent monochromaticity. The advantages of photonuclear reactions using LCS gamma sources for radioisotope production are as follows:

• Differing from bremsstrahlung gamma sources, the LCS gamma sources significantly reduce the heat per useful reaction rate as its γ rays in the energy range of interest are not accompanied by an intense low-energy tail. Moreover, the LCS gamma sources have the ability to selectively tune photons to energies of interest. A high specific activity can be achieved by matching the photon energy to the giant dipole resonance (GDR) peak.

• Differing from the charged-particle induced reactions, the photonuclear reactions generate less heat as their energy deposition in targets via gamma-matter interactions is significantly smaller. Thus, the target can be thicker and requires considerably less cooling [9]. Moreover, the photonuclear reactions have the capability to simultaneously irradiate multiple targets. In this manner, it can maximize the utilization of γ beams and produce multiple radioisotopes simultaneously [8].

The Shanghai Laser Electron Gamma Source (SLEGS) employs the LCS technique. It can generate γ beams ranging from 0.25–21.7 MeV with a full-spectrum flux of 10⁵–10⁷ s^-1 [18] and the best possible bandwidth of γ beams of 5%–15% after passing through a dual collimation system [19]. Using collimation technology, the size of an γ beam can be continuously adjusted to within Φ25 mm [20]. The monochromaticity and high intensity of γ beams combined with detector spectrometers can be used to measure the photonuclear reaction cross-section [21-24]. One of the main topics of SLEGS is the measurement of the key photonuclear reaction cross-section relevant for medical radioisotope investigations, providing crucial nuclear data for the photonuclear method.

In this study, we aim to estimate the production of medical radioisotopes in the energy range of 0.25–21.7 MeV in the SLEGS domain. The remainder of this paper is organized as follows. In Sect. 2, we summarize the radioisotopes that can be produced via (γ, n), (γ, p), and (γ,γ’) reactions and discuss their experimental feasibility. Owing to the scarcity of cross-sectional data regarding photonuclear reactions, theoretical calculations are often used to assess the yields of medical radioisotopes. In Sect. 3, the cross-sectional data of the ¹⁰⁰Mo(γ, n), ⁶⁵Cu(γ, n), and ⁶⁸Zn(γ, p) reactions in the SLEGS energy region were investigated to produce ⁹⁹Mo, ⁶⁴Cu, and ⁶⁷Cu radioisotopes. Finally, we provide a summary in Sect. 4.

Medical radioisotopes produced in photonuclear reactions

In this section, we discuss potential γ, n, (γ, p), and (γ, γ’) reactions for the production of medical radioisotopes. Crucial considerations include the natural abundance of the target isotopes, half-life of the radioisotopes, and available experimental data. Generally, it is preferable to have a target isotope with a dominant abundance (e.g., greater than 10%), and a suitable half-life (several tens of minutes to days) for the produced radioisotope. However, the lack of experimental data may limit the medical applications of radioisotopes.

Different methods are available for measuring photonuclear reactions, including in-beam and offline measurements. Four π ³He neutron detection arrays, charged particle detectors, and nuclear resonance fluorescence spectrometers are commonly used for in-beam measurements of γ, n, (γ, p), and (γ, γ’) reactions. Offline measurement is suitable for determining all three types of photonuclear reactions only in cases where the residual decay has an unambiguous γ-ray and the lifetime of the residual nucleus is within a range of minutes to hours.

Evaluation of cross-sections for ⁹⁹Mo, ⁶⁴Cu, and ⁶⁷Cu production

The cross-section data were used to calculate the expected yield of the radioisotopes for a given thickness and enrichment of the target material, and to determine the optimum energy range for the production of the desired radioisotope and the level of radioisotopic impurities. However, the slow development of gamma sources and small cross-sections of photonuclear reactions resulted in a relatively small number of experimental studies regarding this cross-sectional data. There is often scattering between different experimental datasets. As a result, the cross-sections predicted by the theoretical models play a key role in the production of medical radioisotopes.

The TALYS code [41] offers a unified approach for calculating nuclear reactions that involve neutrons, photons, protons, deuterons, tritons, ³He, and α particles in the keV-200 MeV energy range and for target nuclides with masses of 5 and 339. The code outputs a set of reaction data, for example cross-sections, energy spectra, and angular distributions of the emitted particles, etc.. Numerous studies have tested the TALYS code and demonstrated that it has reliable predictive power in terms of the nuclear reaction calculations [42-44]. In the TALYS code, the decay of the compound nuclear state from the photonuclear reaction was treated using the Hauser-Fenshbach (HF) statistical model [45]. The nuclear-level density and γ strength function were the main input parameters. In this study, we adapted the TALYS-1.96 code. It implemented six different nuclear level density (NLD) models and nine different models of γ strength functions (γSF), as shown in Table 4. The NLD and γSF models employ phenomenological and microscopic approaches, respectively.

In this section, we select the ⁹⁹Mo, ⁶⁴Cu, and ⁶⁷Cu radioisotopes as examples to compare the model calculations with the experimental data. ^99mTc decayed by ⁹⁹Mo is the most frequently used radioisotope in nuclear medicine. ⁶⁸Zn is of interest owing to its applications in the production of the medical radioisotope ⁶⁷Cu; ⁶⁷Cu and ⁶⁴Cu can form a “matched pair" [46, 47]. Therapeutic ⁶⁷Cu, along with positron-emitting ⁶⁴Cu, can measure the uptake kinetics in an organ of a patient by PET imaging, allowing for a precise dosimetric calculation. We used the default values of NLD and the γSF models in the following calculations.

3.1

⁹⁹Mo/^99mTc

With half-lives of 6 h and 140 keV γ-rays, ^99mTc is nearly ideal for SPECT imaging. The most common method to obtain ^99mTc is by elution from ⁹⁹Mo generators. The ¹⁰⁰Moγ, n reaction has a sufficient potential to produce ⁹⁹Mo. The excitation functions are presented in Fig. 1, which covers the entire GDR energy range. The cross-sections of ¹⁰⁰Moγ, n⁹⁹Mo were experimentally measured by Utsunomiya et al. [48], Crasta et al. [49], and Ejiri et al. [14]. The gamma sources used in the experiments were LCS and bremsstrahlung. All the experimental results were distributed in the low-energy region of the GDR and did not exceed the peak. Fig. 1(a) demonstrates that the shape and value of the excitation function are highly sensitive to the choice of the γSF model, particularly at the peak position of GDR. Excluding strength 2 and 3, the remaining γSF models yield relatively similar results that are close to the experimental results. Different γSF models produce similar predictions for energies above the GDR peak. Overall, the cross-sections estimated using Strength 8 achieve the best agreement with the experimental data. As shown in Fig. 1(b), the calculation results of all the NLD models below the peak are consistent but significantly higher than the experimental data, whereas those above the peak are significantly different. The data predicted by the models at energies above 14 MeV are highly valuable. New experiments can be conducted based on SLEGS to verify the model predictions.

Fig. 1Full-screen View

(Color online) ¹⁰⁰Mo(γ, n)⁹⁹Mo reaction cross sections.

3.2

⁶⁴Cu

As listed in Table 1, the decay characteristics of ⁶⁴Cu render it useful for nuclear medicine [50]. It combines PET diagnostic capabilities with those of radiotherapy, with average electron emissions of 190 keV. Currently, ⁶⁴Cu is mainly produced by small cyclotrons through ⁶⁴Ni(p,n) reactions [51-54]. Alternative production using ⁶⁵Cuγ, n does not require rare or expensive ⁶⁴Ni targets and simplifies the chemical separation step. The experimental data for the cross-sections of the ⁶⁵Cuγ, n⁶⁴Cu reaction are presented in Fig. 2 along with the TALYS. The measurements by Katz et al. [55] and Antonov et al. [56] were performed using bremsstrahlung gamma sources. However, they differed significantly in terms of their shape and value. Coote et al. [57] obtained a single cross-section at 17.6 MeV using monochromatic γ-rays from the ⁷Li(p, γ)⁸Be reaction. The cross-section is consistent with the calculation but lower than other experimental data by 80 to 120 mb. The cross-sections of ⁶⁵Cuγ, n⁶⁴Cu can significantly vary with the choice of the γSF models but not the NLD models. However, the discrepancies between the experiments and models cannot be compensated for by varying the γSF models. Future experiments regarding the ⁶⁵Cu photoneutron reaction using a monochromatic LCS gamma source will be useful for resolving these discrepancies and provide assurance for its medical applications.

Fig. 2Full-screen View

(Color online) ⁶⁵Cu(γ, n)⁶⁴Cu reaction cross sections.

3.3

⁶⁷Cu

⁶⁷Cu is a promising β^--particle emitter with an average energy of 141 keV for the targeted radiotherapies. The size of these particles in the tissue was of the same order as the cell diameter. This reduces the unwanted dose burden on patients [58]. With a half-life of 61.83 h and low energy γ emissions (91.266 keV, 7%; 93.311 keV, 16.1%; 184.577 keV, 48.7%), ⁶⁷Cu can provide a long therapeutic effect. ⁶⁷Cu and its stable daughter ⁶⁷Zn are nontoxic to the body, and both Cu and Zn are prevalent trace elements in the body. Along with the PET imaging radioisotope ⁶⁴Cu, it forms a “matched pair”. However, the widespread clinical use of ⁶⁷Cu-based radiopharmaceuticals has been limited by their availability, quantity, and quality. Experiments [59-63] have shown that the ⁶⁸Zn(γ, p) reaction has the potential to produce sufficient quantities of ⁶⁷Cu with a sufficient purity for medical use. Moreover, various Cu and Zn separation methods have been employed in radiochemical processing [34, 64-66].

Figure 3 presents the excitation function of the ⁶⁸Zn(γ, p)⁶⁷Cu reaction. Experimental data regarding this reaction are scarce, with only one group of two points in the energy region of SLEGS [67]. The data are significantly inconsistent with any calculation and hardly constrain NLD and γSF models. In contrast to the γ, n reaction of several hundred mb, the (γ, p) reaction is typically several mb in the medium-heavy nuclei. Additionally, protons can easily stop at the target and are difficult to detect. If the product nuclei of the (γ, p) reaction are unstable and have a moderate half-life, the cross-sections can be determined by offline measurements of the characteristic γ-rays during decay. For the ⁶⁸Zn(γ, p) reaction, an offline measurement technique can be employed to determine its cross-sections.

Fig. 3Full-screen View

(Color online) ⁶⁸Zn(γ, p)⁶⁷Cu reaction cross sections.

Summary

In this study, we summarize the medical radioisotopes that can be produced by γ, n, (γ, p), and (γ, γ′) reactions, along with the experimental data for these production pathways. We investigated the photonuclear reaction cross-sections for the production of ⁹⁹Mo, ⁶⁴Cu, and ⁶⁷Cu. The experimental data for ⁹⁹Mo and ⁶⁷Cu did not cover the entire GDR energy region, whereas the data for ⁶⁴Cu exhibited a significant discrepancy. To better constrain the NLD and γSF models in TALYS, additional experimental measurements at SLEGS in the energy region of GDR are necessary in the future.

To compare the productivity of the photonuclear method with that of traditional methods, we consider the production of ⁹⁹Mo/^99mTc as an example. The highest flux currently available for SLEGS is 10⁷ s^-1. We adopted the calculated results by the TALYS-1.96 code using the Strength 8 model for the yield calculations, which provided a maximum of 145 mb of the ¹⁰⁰Mo(γ, n) reaction cross-section at the incident photon energy of 14.4 MeV. After irradiating a ¹⁰⁰MoO₃ target with a radius of 2 mm and a thickness of 1 cm for 5 times the half-life of ⁹⁹Mo, the saturation specific activity of ⁹⁹Mo and ^99mTc can reach 4.76 and 4.74 μ Ci/g, respectively. The use of stacked targets can further enhance production. When the energy regions corresponding to the maximum cross-sections of the targeted nuclear reactions are close, multiple radioisotopes can be simultaneously produced. For example, the ⁶²Cu, ⁶⁴Cu, and ⁸⁹Zr radioisotopes can be produced simultaneously when the incident photon energy is approximately 17 MeV.

Currently, ⁹⁹Mo is primarily produced by the (n,f) reaction in a high-flux reactor using an enriched ²³⁵U target. The specific activity of ⁹⁹Mo can reach 185 TBq/g (5000 Ci/g) [6]. However, most reactors will gradually shut down by 2030 [68]. Medical cyclotrons can be used to directly generate ^99mTc via the ¹⁰⁰Mo(p,2n) reaction. According to the experimental data, an enriched ¹⁰⁰Mo target irradiated with a 16.5 MeV proton beam at 130 μA for 6 h yielded 116 GBq/g (3.13 Ci/g) of ^99mTc [69]. Recently, a subcritical ⁹⁹Mo production system was developed, which is driven by an accelerator-based deuterium–deuterium (D–D) neutron source. The D–D fusion reaction generates neutrons that irradiate a low-enriched uranium solution and induce fission in ²³⁵U. The system can generate 47.8 mCi/g ⁹⁹Mo for a stable 24 h operation with a neutron intensity of 1×10¹⁴ n/s [70]. Bremsstrahlung gamma sources based on linear electron accelerators are often used to produce ⁹⁹Mo radioisotopes. Using a ¹⁰⁰MoO₃ target irradiate with a 35 MeV electron beam at 100 μA for 20 h; the specific activity can achieve 4.4 GBq/g (119 mCi/g) [71]. NorthStar Medical Radioisotopes developed a ⁹⁹Mo production system using an electron linear accelerator. Using this system based on γ, n reactions, approximately 30% more ⁹⁹Mo is produced per gram of the target material compared with the traditional neutron capture route [72, 73].

In comparison, the specific activity produced by the photonuclear method based on SLEGS was low. With a further increase in the flux of γ beams, from 10⁷ s^-1 to 10¹⁵ s^-1 [20], the specific activity of ⁹⁹Mo can reach up to 4×10² Ci/g. For the 2-day protocol, the activity of the ^99mTc-labelled tracers required for one myocardial perfusion imaging was 24 mCi [74]. The total yield of stacked targets in one year can provide 400,000 myocardial perfusion images. As the flux of SLEGS increases in the future, this method is promising for producing medical radioisotopes and will become feasible in China.