Chemical Purification of Terbium-155 from Pseudo-Isobaric Impurities in a Mass Separated Source Produced at CERN

Four terbium radioisotopes (149, 152, 155, 161Tb) constitute a potential theranostic quartet for cancer treatment but require any derived radiopharmaceutical to be essentially free of impurities. Terbium-155 prepared by proton irradiation and on-line mass separation at the CERN-ISOLDE and CERN-MEDICIS facilities contains radioactive 139Ce16O and also zinc or gold, depending on the catcher foil used. A method using ion-exchange and extraction chromatography resins in two column separation steps has been developed to isolate 155Tb with a chemical yield of ≥95% and radionuclidic purity ≥99.9%. Conversion of terbium into a form suitable for chelation to targeting molecules in diagnostic nuclear medicine is presented. The resulting 155Tb preparations are suitable for the determination of absolute activity, SPECT phantom imaging studies and pre-clinical trials.

They exist predominately in the III+ oxidation state under aqueous conditions. The exceptions are europium, which can be selectively reduced to Eu(II) under strongly reducing conditions, and cerium, which can be easily oxidised to Ce(IV). Changes in oxidation state markedly influence chromatographic behaviour and this can be exploited when developing separation methods.
A well-known method of separating lanthanide elements utilises cation-exchange chromatography with α-hydroxyisobutyric acid (α-HIBA) eluent and provides good separation even from neighbouring elements 1,5,8 . However, the process is slow and requires precise control of chemical conditions (pH and α-HIBA concentration) to give optimal yield and purity. Attempts to accelerate separation tend to compromise terbium recovery.
A significant 139 Ce (T 1/2 = 137.6 d 17 ) impurity exists in 155 Tb sources from CERN-ISOLDE and CERN-MEDICIS owing to formation of the pseudo-isobaric species, 139 Ce 16 O, which cannot be removed by mass separation. Given its half-life, it constitutes an increasing proportion of overall source activity during transport and storage. In this study, we present a simple method for producing radiologically pure terbium preparations in a chemical form suitable for chelation to targeting molecules as well as for absolute activity measurements and phantom imaging studies. Our aim was to develop a robust, efficient and rapid method capable of isolating terbium from the foil matrix as well as from 139 Ce by selective oxidation. Therefore, ion-exchange and extraction chromatography resins were chosen based on their selectivity for tetravalent over trivalent species.

Results
Chemical separation. Batch separation. In the presence of an oxidant (sodium bromate, NaBrO 3 ) and in HNO 3 solutions commercial UTEVA, TEVA and TK100 extraction resins (Triskem International) and AG1 anion exchange resin (BioRad) all showed pronounced cerium adsorption selectivity over terbium (Fig. 1). The results imply oxidation of cerium to Ce(IV) was achieved, with terbium remaining in the trivalent state (Tb(III)).
High Ce adsorption (K d = 100-1,000) was observed at high HNO 3 concentrations (8-10 M) on all four resins, whilst terbium adsorption remained minimal (K d = 0.1-10) across the concentration range (Fig. 2). The best separation resolutions (Equation (2), SR > 100) were obtained using TEVA and UTEVA resins at high HNO 3 concentrations; further studies were conducted on these resins using pre-packed cartridges.
Kinetic studies. UTEVA extraction chromatography resin was chosen to demonstrate kinetic behaviour with the rate of cerium adsorption studied in 10 M HNO 3 /0.1 M NaBrO 3 solutions; rapid adsorption (<60 s) was observed (Fig. 3a). The rate of cerium oxidation was also studied in 10 M HNO 3 /0.1 M NaBrO 3 solutions. Solutions were filtered under vacuum after a minimum of 60 s in contact with the resin. Rapid oxidation (<90 s) of cerium was observed (Fig. 3b).
Neither the rate of adsorption nor the rate of oxidation were limiting factors in the separation, suggesting that rapid column separation is achievable under these conditions. www.nature.com/scientificreports www.nature.com/scientificreports/ Column studies. Column-based separation using a commercially available pre-packed UTEVA cartridge (2 mL) provided effective isolation of terbium from cerium impurities. The elution profile (Fig. 4) shows that terbium (>99%) was removed in the load solution (10 mL, 8 M HNO 3 ) and the subsequent wash solution (10 mL, 8 M HNO 3 ) with minimal cerium impurities remaining (<0.002%). Cerium was successfully recovered by elution from the cartridge in hydrochloric solution (<10 mL, 0.1 M). The column-based separation was repeated using a pre-packed TEVA cartridge (2 mL); however, the separation achieved was less successful as ~0.1% Ce was detected in the Tb fraction under similar conditions (Fig. 4).

Discussion
In many cases, it is essential that suitable radiochemical methods are available to provide radionuclides in sufficient quantities with relatively high specific activity, radionuclidic and chemical purity to facilitate accurate pre-clinical and clinical study. The method described is able to produce high radiological purity 155 Tb sources, suitable for absolute activity, nuclear data and ionisation chamber measurements. The sources are also suitable for bioconjugation, molecular chelation and SPECT imaging studies. Although the 139 Ce impurity discussed here does not possess significant biological toxicity 18 , it is radioactive and, if not removed, would result in an unnecessary additional dose to the patient.
Currently, proton-induced spallation is the main route for producing 155 Tb at CERN for (pre)-clinical studies. The chemical purification method proposed here (Fig. 5) allows the quantitative separation of 155 Tb from a zinc and/or gold matrix and from 139 Ce impurities produced by spallation at the CERN-ISOLDE and CERN-MEDICIS facilities. The method is rapid, simple and can also be used to recover a high purity 139 Ce source; a useful standard in gamma spectrometry (E γ = 165.86 keV, 79.90%) 17 .
The method has not yet been validated for the removal of other stable (e.g. 139 La 16 O + , 155 Gd + ) or longer-lived, radioactive (e.g. 155 Eu + ) isobaric impurities; as with 139 Ce 16 O, they would not be removed by mass separation. Such impurities might not pose a significant toxicological risk if they were to enter the body 19,20 but nevertheless, would form stable complexes with DOTA (logK > 22) 21 and DOTA-containing targeting molecules 4 and could compete with the target terbium isotope(s), reducing their efficacy.     www.nature.com/scientificreports www.nature.com/scientificreports/ additional terbium-specific tuning was carried out. A 209 Bi (10 ppb solution in 2% v/v HNO 3 ) internal standard was used to monitor and correct for instrumental drift during longer runs. Blank HNO 3 (2% v/v) solutions were monitored regularly to ensure no Ce or Tb cross-contamination during a run.
Gamma-ray spectrometry. An n-type HPGe γ-ray spectrometer with a resolution (FWHM) of 1.79 keV at 1.33 MeV and relative efficiency 28% was used to determine the 139 Ce/ 155 Tb activity ratio. The detection system set-up and full-energy peak efficiency calibration is described in detail by Collins et al. 22 .
The nuclear data (half-lives and γ-ray emission intensities) used to determine the activities of 155 Tb and 139 Ce were taken from the evaluated database of ENSDF and the DDEP, respectively 3,17 . As 139 Ce could not be observed after the chemical separation, the activity ratio of the 139 Ce/ 155 Tb in the chemically separated solution was estimated from the detection limit of the detector for 139 Ce 23 .

Irradiation conditions and mass separation.
Terbium-155 sources used in this study were produced at the CERN-ISOLDE and CERN-MEDICIS facilities. Three 155 Tb sources were produced and provided to NPL for chemical separation between 2017 and 2018. The irradiation conducted at CERN-MEDICIS was as follows: A high purity Ta metal target (Ta647M) made of 12 rolls of Ta foil (99.95% purity, 12 μm thick, 15 mm wide, 2 cm diameter) with a total mass of 357 g was arranged in a 20 cm long Ta tube coupled to a rhenium surface ion source. The target was irradiated with 1.4 GeV protons delivered by the Proton Synchrotron Booster accelerator (CERN, Geneva). The CERN-MEDICIS irradiation target is located in the High Resolution Separator (HRS) beam dump position at ISOLDE (Fig. 6), and receives a fraction of the scattered 1.8 × 10 18 protons downstream from a primary HRS target (623SiC, ISOLDE physics program). The irradiation was scheduled within the MED004 approved experiment and took place from 27 th September to 1 st October 2018. The irradiated target was then moved to the CERN-MEDICIS isotope mass separator in order to release and extract ion species selected at mass-to-charge ratio of 155 6 . The separated ions were collected on a zinc-plated gold foil and removed on 3 rd October. The following isotopes were implanted upon sample retrieval: 139 Ce (implanted as 139 Ce 16 25,26 showing the incoming proton beam on an ISOLDE target (3.5 g/cm 2 UC x for the purpose of the simulation) and intercepting the MEDICIS target downstream. Middle: Screenshot taken with the beam scanner, located before the implantation chamber. Beams at A/q = 154,155,156 are seen (153, 157 partly visible). The collected beam is centred on A/q = 155, while isotopes present at other masses are physically removed from the implantation using mechanical slits located ahead of the foil. The horizontal scale is in mm. Bottom: Two zinc-coated gold foils in the collection chamber seen from the rear. The collection takes place on the foil located on the left. www.nature.com/scientificreports www.nature.com/scientificreports/ remaining solution was added to 0.1 g of resin (UTEVA, TEVA, TK100 or AG1). Sodium bromate (0.1 M, 0.03 g) was added to identical samples to assess changes in adsorption to the resin as a result of selective oxidation of Ce. In all cases, the samples were shaken and left to equilibrate for 24 h. After equilibration, the solutions were filtered to isolate the aqueous phase (Whatman 41 ashless filter paper, 20-25 μm pore size). An aliquot was taken from each sample, diluted with 2% HNO 3 (2% v/v) and analysed by ICP-MS.
The adsorption of Tb and Ce onto each resin was quantified by calculating the distribution coefficient (K d ) using Eq. (1) 24 Where (CPS) 0 and (CPS) t are the concentrations of analyte in the aqueous phase before and after equilibration, respectively, V is the volume of solution (mL) and m is the mass of resin used (g). The separation achievable in the different HNO 3 solutions was quantified by calculating the separation factor using Eq. (2). Column studies. Column-based separation was studied using a pre-packed 2 mL UTEVA cartridge (50-100 µm, Triskem International). The resin was pre-conditioned with 8 M HNO 3 (20 mL). A HNO 3 solution (10 mL, 8 M) containing 0.1 M NaBrO 3 , 100 ppb Tb and 100 ppb Ce was loaded onto the resin. A wash solution of 10 mL 8 M HNO 3 was added to ensure removal of all Tb from the cartridge. Subsequent elution of Ce was achieved using 20 mL 0.1 M HCl. This separation method was also repeated using a pre-packed 2 mL TEVA cartridge (50-100 µm, Triskem International).
Throughout the separations, 1 mL fractions were collected, diluted with HNO 3 (2% v/v) and analysed by ICP-MS in order to compile an elution profile. Column separations were carried out under gravity (approximate flow rate = 0.3 mL/min).
Method validation with active sample. Three zinc-coated gold foils containing 155 Tb and 139 Ce were received at NPL from CERN-ISOLDE and CERN-MEDICIS. The radionuclides were leached by dissolving the zinc layer in 20 mL 6 M HCl and the gold foil in 20 mL aqua regia. Both layers were dissolved in order to maximise the yield of terbium from the sources received. The combined solution was evaporated gently on a hot plate (~150 °C) to incipient dryness and re-dissolved in a 10 mL 8 M HNO 3 /0.1 M NaBrO 3 solution. An ampoule was prepared for HPGe gamma spectrometry in order to quantify the activity of 155 Tb and 139 Ce present. After analysis, the portion was recombined with the bulk solution.
A pre-packed 2 mL UTEVA cartridge (Triskem International, 50-100 µm) was conditioned with 20 mL 8 M HNO 3 . The 10 mL sample was then loaded onto the column and the fraction collected under gravity. The column was washed with 10 mL 8 M HNO 3 . This fraction was collected, under gravity, and combined with the load fraction. The combined fractions were evaporated gently on a hot plate (~150 °C) to incipient dryness and re-dissolved in 20 mL 0.1 M HCl.
An ampoule of the combined terbium fractions was prepared and analysed by HPGe gamma spectrometry in order to assess the resultant purity of the 155 Tb source after separation. The terbium recovery was calculated as follows: (1 ) where R 0,1 and R 0,2 are the count rates of the 105 keV gamma-ray emission before and after separation of the 139 Ce, respectively at the reference time 2017-09-29 12:00 UTC. N 1 and N 2 are the net peak areas of the 105 keV full-energy peak measured before and after separation, Δt L,1 and Δt L,1 are the measurement live times, m 1 and m 2 are the measured active masses of solution, m D and m E are the total mass of solution used to dissolve the Zn layer of the target and eluent used in the chemical separation, respectively, λ is the decay constant of 155 Tb, t 1 and t 2 are the time elapsed since the reference time and Δt 1 and Δt 2 are the measurement real times.

Conclusion
A novel method has been developed for the isolation of 155 Tb from sources produced at CERN-ISOLDE and CERN-MEDICIS, currently the main producers of the isotope. A high purity 155 Tb preparation was successfully recovered from a zinc-coated gold matrix and from 139 Ce impurities using a chromatography-based system. The method was shown to be capable of separating 100 ppb Tb and Ce in a 10 mL solution, equivalent to ~6 GBq 155 Tb and ~0.25 GBq 139 Ce. The radiologically pure 155 Tb preparation was subsequently used for absolute activity measurements and ion chamber measurements. The preparations are also suitable for phantom imaging and pre-clinical studies.

Data Availability
The data generated and analysed during this study are available, upon reasonable request, from the corresponding author.