Introduction

There is a growing interest in the possible practical applications of secondary neutrons generated on medical radioisotope production targets during routine bombardments. High-intensity target stations are becoming more abundant, with beam currents up to 1 mA in some cases. An intense source of energetic neutrons close to the target surface can thus be obtained. Preliminary results have already proved that secondary neutrons generated at intermediate energy accelerator facilities (Ep > 60 MeV) have promising potential for radioisotope production on a scale that can support various research initiatives (cf. Engle et al. [1, 2]). The list of possible candidates [2] is rather attractive: 36Cl, 47Sc, 63Ni, 64Cu, 67Cu, 85Kr, 89Zr, 224Ra, 225Ac, 229Th, 231Pa and 237Np.

Recently, Auditore and co-workers [3] investigated the production possibility of 64Cu with secondary neutrons coming from an [18O]-H2O target (for 18F production) located at a lower-energy cyclotron (Ep = 30 MeV). According to their theoretical study, a yield of about 457 MBq (12.3 mCi) is expected on a highly enriched 64Zn target (via the 64Zn(n,p)64Cu nuclear reaction) using a 120 μA beam and 2 h bombardment. Their encouraging results prompted us to study the production possibility of other medically relevant radioisotopes using secondary neutrons generated by protons, even from a compact medical accelerator (Ep < 20 MeV).

As a part of this study, we have first investigated the formation of the therapeutically relevant 161Tb (T1/2 = 6.89 d) via the 160Gd(n,γ)161Gd → 161Tb nuclear process. Due to its unique decay mode (i.e. a low-energy β− emitter, but emits also a significant amount of conversion and Auger electrons), it provides a better therapeutic effect on tumors than the widely used 177Lu (T1/2 = 6.647 d). Additionally, its pathway inside the human body can be monitored via a gamma-camera since it also emits low-energy photons (see e.g. Lehenberger et al. [4]). It should be also noted that element of Terbium is also a good candidate for theranostics. This consists of the use of a positron emitter of an element together with a therapeutic radionuclide of the same element. This way the diagnosis and therapy are combined and it becomes a personalized medicine. Both the 161Tb(Therapy)/152Tb(PET) and 149Tb(Therapy)/152Tb(PET) pairs can be used for this purpose.

Nuclear reactors are typically used for routine production of 161Tb by irradiating massive and highly enriched 160Gd targets [4]. Due to the differences of the neutron flux of a reactor (dominantly thermal neutrons) and that emitted from a ‘radioisotope production target’ (dominantly energetic neutrons), a much lower yield is expected for 161Gd in the latter case. However, it could be enough for in-house labeling experiments. We report here our preliminary results on the formation of 161Tb via secondary neutrons using the 9Be + p neutron source of the MGC-20E cyclotron of ATOMKI (Debrecen, Hungary).

Experimental

Two sample discs were prepared by compaction of Gd2O3 powder of natural isotopic composition using a hand press tool. These samples had the same diameter (1 cm) but their masses were slightly different: 328.4 mg (Sample 1) and 339.9 mg (Sample 2). Sample 1 was enclosed into a Cd case of 1 mm wall thickness. The Cd case fully shielded Sample 1 against thermal neutrons. In contrast, Sample 2 was exposed to the full neutron spectrum. Both targets were placed into thin polyethylene (PE) bags before being mounted onto a 9 cm × 9 cm × 0.1 cm holder plate made of AlMgSi alloy. For monitoring the neutron fluence, a rectangular Al plate (dimensions 16 mm × 16 mm × 1.1 mm, 99.99% chemical purity) was also mounted onto the back side of the AlMgSi holder plate. This assembly was installed at the irradiation position. Figure 1 shows a schematic diagram of the irradiation arrangement. This arrangement ensured that the expected neutron spectra at the positions of Samples 1 and 2 were practically identical because of the cylindrical symmetry of the irradiation field around the θ = 0° direction (the direction of the bombarding beam).

Fig. 1
figure 1

Schematic diagram of the 9Be + p neutron source and the target holder used for the 161Tb production experiment

Full size image

Originally, we wanted to reproduce the irradiation conditions of Auditore et al. [3], i.e. using a Be energy degrader foil in front of an [18O]-H2O target. However, a preliminary investigation made it clear that the neutron fluence rate from such a target at our lower proton-beam energy of 18 MeV will be much lower than in the case of the 30 MeV beam used by Auditore and co-workers. We considered this neutron fluence rate to be too low for practical purposes. Additionally, due to the lower incident proton energy we could not use Be or any other energy degrader without significantly decreasing the available 18F yield. Moreover, without a Be degrader the fluence rate of secondary neutrons will be almost an order of magnitude lower, according to the calculations of Auditore and co-workers. Thus, in order to keep the irradiation time within acceptable limits in our initial investigation, a Be target was used instead of the [18O]-H2O target. For this purpose, the existing 9Be + p neutron irradiation facility (Fenyvesi [5]) operated by the MGC-20E cyclotron facility was employed. A 3 mm thick stopping beryllium target was used for producing Be + p neutrons. The bombarding protons impinged on the target with an energy of Ep = 17.8 MeV ± 0.6%. The beam current was Ip = 15.31 µA. Sample 1 and Sample 2 were exposed to the 9Be + p neutrons for an irradiation time of 32656 s. The total collected charge reached 0.5 C. Under these conditions, a neutron fluence of 1.0 × 1014 cm−2 ± 5% could be achieved at the irradiation position, 79 mm downstream from the back surface of the Be target (see Fig. 1).

The activated samples were counted separately, first non-destructively using a well-calibrated HPGe detector (Canberra) that was coupled to a 4096-channel MCA (Nucleus). Note that Sample 1 was counted after removal from the Cd case. Spectra of the gamma photons emitted by the samples were collected several times. The detection efficiency as a function of the gamma energy was measured via counting point-like calibrated gamma sources of known activity. The full energy photo-peaks identified in the spectra were evaluated using the built-in software of the MCA. The obtained net peak areas were used for the activity calculations.

After completion of the non-destructive counting of the samples, both samples were dissolved and the Tb isotopes were separated from the Gd matrix (Brezovcsik et al. [6]). All fractions were counted separately.

The uncertainties of the measured activities were estimated in a standard way according to the well-known JCGM 100:2008 document [7]. The independent relative errors of the linearly contributing processes were summed in quadrature. These were integrated charge (8%), target mass (1%), decay data (3%), detector efficiency (5%) and peak area determination (1–7%). The overall error was around 13%.

Monte Carlo calculations

Neutron transport simulations with Release 1.60 of the MCNP5 code [8] were performed for obtaining the neutron spectrum at the positions where the gadolinium sample discs were irradiated. The Al fluence monitor sample, both natGd2O3 discs, a model of the target holder and the concrete shielding walls of the irradiation vault were implemented on the geometry cards of the MCNP input. For the source term, a uniformly sampled volume source was defined that emitted neutrons with realistic energy and angular distributions, as described below.

The energy distribution of the yield of the Be + p neutrons at Ep = 17.815 MeV was obtained via interpolation using the data published by Brede et al. [9]. The angular distribution was derived from the data of Brede et al. [9] and Lone et al. [10].

The xsdir_mcnp5_7.1 cross section data library was used for the transport calculations. This library consists of the ENDFB-7.1 cross section data library and the discrete S(α,β) data for thermal scattering taken from the ENDFB-7.0 library.

Cell averaged F4:n tallies were used for calculating the neutron fluences for the Al fluence monitor and the natGd2O3 discs. FM:4 tally multiplier cards were used for estimating the saturation activities of 24Na, 159Gd and 161Gd produced via the 27Al(n,α)24Na, 160Gd(n,x)159Gd and 160Gd(n,γ)161Gd nuclear reactions, respectively. The cross section data for the tally multiplier card used in the case of the 160Gd(n,γ)161Gd and natGd(n,x)159Gd reactions were taken from the TENDL-2017 library [11]. Spectral neutron flux tallies obtained with MCNP5 for the Gd2O3 samples and the Al flux monitor; and the spectral neutron flux rate for the neutrons from the 9Be + p source can be seen in Fig. 2.

Fig. 2
figure 2

Spectral neutron flux tallies obtained with MCNP5 for the Gd2O3 samples and the Al flux monitor; and the spectral neutron flux rate for the neutrons from the 9Be + p source

Full size image

Results

The measured 24Na activity of the Al neutron fluence monitor was AEOB = 1.45 × 105 Bq at the end of the irradiation, corresponding to Asat = 4.24 × 105 Bq saturation activity. In the measured gamma-ray spectra we could identify only two radioisotopes, namely 159Gd (T1/2 = 18.479 h) and 161Tb. However, the formation of 156,157Eu and 151,153Gd is also possible, taking into account their threshold energies and the excitation functions of their contributing reactions on 157,158Gd target nuclei (see e.g. the TENDL-2017 [11] and EXFOR [12] databases). Unfortunately, the produced activities of those radioisotopes were well below the detection level of our counting set-up. The decay data of the measured radioisotopes were taken from the NuDat 2.7 Interactive Chart of Nuclides [13]. The relevant decay data, reactions and Q-values are listed in Table 1.

Table 1 Decay data and relevant reactions of the investigated reaction products
Full size table

161Tb formation

In the case of a gadolinium target, neutrons can produce only the precursor of 161Tb (see Table 1). This radioisotope (161Gd) has a relative short half-life (T1/2 = 3.7 min) and decays completely via β− to the ground state of 161Tb. (The 161Tb activity measurements therefore started after the complete decay of 161Gd.)

Only one reaction forms 161Gd on a natural target: 160Gd(n,γ)161Gd. Some measured cross-section data for the above reaction are shown in Fig. 3 (Kononov et al. [14], Perkin et al. [15], Sahota et al. [16], Singh et al. [17], Valkonen et al. [18], Voigner et al. [19], Wille and Fink [20]). These experimental results are compared in Fig. 3 with the evaluated theoretical prediction by means of the TALYS code, as compiled in the TENDL-2017 library [11].

Fig. 3
figure 3

Excitation function for the 160Gd(n,γ)161Gd nuclear reaction

Full size image

It can be seen that the calculated curve shows rather good agreement with the experimental values up to about 4 MeV. Unfortunately, no detailed measurements are available beyond this energy. According to the TENDL-2017 results, the excitation function curve of the 160Gd(n,γ)161Gd reaction decreases almost monotonically as a function of increasing energy and has a minimum at about 7 MeV. Above this energy it has a local peak of 1 mb at about 12 MeV. For the yield calculations, these TENDL-2017 values were used.

The measured EOB (End Of Bombardment) activity of the Cd-covered Sample-1 and Sample-2 (which was irradiated without Cd shielding) are listed in Table 2. Table 2 also contains the predicted 161Tb yields based on the Monte Carlo calculations. The expected experimental yields of 161Tb, normalized to enriched 160Gd (100%) targets, are about 8000 and 7240 Bq C−1 g−1, respectively. The higher yield in the case of the shielded sample was expected since the Cd wrap could slow down slightly the energetic neutrons. The Monte Carlo calculations also support this difference however overestimate the experimental results.

Table 2 The EOB activities obtained from the measurements and simulations
Full size table

159Gd formation

Two reactions populate the ground state of 159Gd directly, namely 158Gd(n,γ)159Gd and 160Gd(n,2n)159Gd in the case of natural Gd target. This ground state decays to 159Tb, which is stable. Unfortunately, no cross-section values were found in the literature for the first reaction, while the second one has only one detailed measurements in the investigated energy region. We have therefore created the excitation function curve for the natGd(n,x)159Gd nuclear process using the TALYS predictions for the two separate reactions (TENDL-2017). The excitation function curve seems to be similar to the 160Gd(n,γ)161Gd reaction (see Fig. 4). However, it has a much sharper increase in the case of the peak and its maximum is two orders of magnitude higher (about 400 mb) [21]. The measured and calculated EOB activities are listed in Table 2. For 159Gd, the Monte Carlo calculations underestimate the experimental results. The expected yields for natural metallic target are 35.14 and 28.67 kBq C−1 g−1. Note that in the case of highly enriched 160Gd target the expected yield will be lower since only the 160Gd(n,2n)159Gd reaction participate in the formation of 159Gd.

Fig. 4
figure 4

Excitation function for the natGd(n,x)159Gd nuclear reaction

Full size image

Conclusions

Based on our preliminary results, it can be concluded that the fluence rates of the secondary neutrons generated on medical radioisotope production targets during routine bombardments at a low energy accelerator could not be enough for practical 161Tb production. However, using an additional neutron source together with the radioisotope target could produce a higher neutron fluence rate for the above purpose. Depending on the incident maximum proton energy and the energy window for the formation of the primary product (cf. 18F, 64Ni, etc.) an energy degrader (Be, Al plates) could be the source of the required neutron excess (Auditore et al. [3]). For protons below 20 MeV when energy degraders cannot be used, a higher neutron fluence rate can be achieved by replacing the beam collimator material with Be and/or simply using only a Be-target after removing the radioisotope production target assembly. Our present irradiation scenario with a separate 9Be-source resulted in a rather low yield for 161Tb (8000 and 7240 Bq C−1 g−1, respectively). It is expected, however, that activation of a highly enriched 160Gd sample (0.1–0.5 g) for 3–4 days with a high beam current (above 100 μA) could produce around 3.7–11.1 MBq (0.1–0.3 mCi) 161Tb at EOB, enough for labeling studies and preclinical experiments.