Separation and recovery of exotic radiolanthanides from irradiated tantalum targets for half-life measurements

The current knowledge of the half-lives (T1/2) of several radiolanthanides is either affected by a high uncertainty or is still awaiting confirmation. The scientific information deriving from this imprecise T1/2 data has a significant impact on a variety of research fields, e.g., astrophysics, fundamental nuclear sciences, and nuclear energy and safety. The main reason for these shortcomings in the nuclear databases is the limited availability of suitable sample material together with the difficulties in performing accurate activity measurements with low uncertainties. In reaction to the urgent need to improve the current nuclear databases, the long-term project “ERAWAST” (Exotic Radionuclides from Accelerator Waste for Science and Technology) was launched at Paul Scherrer Institute (PSI). In this context, we present a wet radiochemical separation procedure for the extraction and purification of dysprosium (Dy), terbium (Tb), gadolinium (Gd), and samarium (Sm) fractions from highly radioactive tantalum specimens, in order to obtain 154Dy, 157-158Tb, 148,150Gd, and 146Sm samples, needed for T1/2 determination studies. Ion-exchange chromatography was successfully applied for the separation of individual lanthanides. All separations were conducted in aqueous phase. The separation process was monitored via γ-spectrometry using suitable radioactive tracers. Both the purity and the quantification of the desired radiolanthanides were assessed by inductively coupled plasma mass spectrometry. Test experiments revealed that, prior to the Dy, Tb, Gd, and Sm separation, the removal of hafnium, lutetium, and barium from the irradiated tantalum material was necessary to minimize the overall dose rate exposure (in the mSv/h range), as well to obtain pure lanthanide fractions. With the herein proposed separation method, exotic 154Dy, 157-158Tb, 148,150Gd, and 146Sm radionuclides were obtained in sufficient amounts and purity for the preparation of samples for envisaged half-life measurements. During the separation process, fractions containing holmium, europium, and promethium radionuclides were collected and stored for further use.


Introduction
Exotic radionuclides with relatively long half-lives are of great interest in a wide range of research domains. Benchmark radionuclides for astrophysics studies are for example the nuclides 146 Sm, 148, 150 Gd,and 154 Dy [1][2][3], whose production (and destruction) pathways provide essential information on the creation and on the development of the known Universe. Long-lived radiolanthanides such as 157 Tb and 158 Tb, produced during irradiation of the urania-gadolinia fuel, are of extreme importance in the field of nuclear energy, since the exact knowledge on the nuclear properties of these nuclides is necessary for nuclear fuel disposal and for nuclear transmutation processes [4]. Furthermore, 157 Tb was lately marked out as a relevant nuclide for the discovery of sterile neutrinos in the keV mass region [5] whereas, in a more futuristic approach, 158 Tb was suggested as a valid decay-energy source in nuclear accumulators [6]. It is apparent that these studies and applications require precise nuclear data, e.g., the half-lives (T 1/2 ), of the above-mentioned long-lived radiolanthanides. However, for 146 Sm, 148, 150 Gd,154 Dy, and 157,158 Tb, only discordant or uncertain information on their T 1/2 is available. As an example, a collection of T 1/2 values for 146 Sm, 154 Dy, and 157 Tb reported in literature is given in Table 1.
Discrepancies in the available T 1/2 data for 150 Gd are observed as well, with values differing from one another of at least 20% [10,24,25]. For 158 Tb, former measurements showed estimated uncertainties of 30% [26] and 20% [27], with the currently adopted T 1/2 = 180 ± 11 y [28] still awaiting an independent corroboration. The scientific impact of such uncertain/contradictory values is considerable. A prominent example is 146 Sm, whose recent T 1/2 re-evaluation might implicate a redetermination of the chronological formation of the Solar System [12]. Currently, a confirmation on this latest measured value is still pending. A similar situation holds for 148 Gd, for which a more recent T 1/2 = 70.9 ± 1.0 y [29] was estimated. This value, Tb 160 y 40 y 1963 [16] 280 y 120 y 1964 [17] 150 y 30 y 1964 [18] 99 y 10 y 1983 [19] 71 y 7 y 1996 [20] c given only as a preliminary result, supports the previous measurement of Prestwood et al. [30]. However, an update on this long-term measurement is still lacking. Reasons for these shortcomings in the nuclear databases are in many cases the limited availability of the isotopes of interest, often accompanied by difficulties in the preparation of samples with a high chemical purity. With the initiative "Exotic Radionuclides from Accelerator Waste for Science and Technology-ERAWAST", launched by Paul Scherrer Institute (PSI) [31,32], these limitations are circumvented by using highly active accelerator waste from the PSI accelerator-facilities as a source of exotic radionuclides for scientific purposes. Detailed information on this longterm project can be found at the website "psi.ch/en/lrc/erawast". PSI owns one of the most powerful accelerators in Europe, i.e., a ring-cyclotron with 590 MeV proton energy and 2.2 mA beam current, which feeds the SINQ neutron spallation source. Between 2000 and 2001, the second SINQ Target Irradiation Program (STIP-II) [33] was performed at PSI, with the purpose of studying radiation damage effects on construction materials induced by highenergy protons and spallation neutrons. During that project, 51 samples of tantalum were exposed to fluxes of high-energy protons and spallation neutrons in the SINQ facility. It is known from previous studies [34][35][36] that the majority of the radionuclides produced in Ta targets irradiated with protons and spallation neutrons are lanthanides (Lns), with isotopic compositions considerably enriched in neutron-deficient nuclides such as 146 Sm, 148,150 Gd,154 Dy, and 157,158 Tb. After a cooling period of almost 20 years, the active Ta material from the STIP-II project represents an extraordinary source of exotic Sm, Gd, Dy, and Tb radionuclides to be used within the scope of ERAWAST. It is important to mention that the method applied for extracting the desired exotic Lns from the radioactive Ta matrix is crucial, since the purity of the retrieved material strongly affects the uncertainty of the half-life determination studies. For the separation of Lns several techniques are known, summarized in [37][38][39]. In particular, the recovery of Lns from active Ta targets was reported in [40][41][42]. However, in those studies (mostly involving amounts of Lns in the order of milligrams), the separation efficiency for each Lns was not stated, and furthermore, the applied procedures required an extensive use of organic solvents. Recently, a method for the extraction of ppb amounts of Dy, Gd, and Sm nuclides from proton irradiated Ta targets was reported for cross section measurements [43]. The separated Dy fraction contained traces of Ho and Tb, while the Sm fraction included small amounts of Eu and Pm. However, for the purpose of that work-the determination of production cross sections-a further separation of the lanthanide fractions was not necessary and, hence, was not performed.
Here, we report a chemical separation system for the extraction of Sm, Gd, Tb, and Dy from irradiated Ta samples, in sufficient amounts and purity for future half-life measurements. The separation process was monitored using γ-spectrometry. The purity as well the isotopic composition of each Lns fraction was assessed by inductively coupled plasma mass spectrometry (ICP-MS).

Materials
Irradiated Ta samples. Depending on their shapes, the Ta specimens were classified as "Tensile", "Bend bar", and "Charpy". For the irradiation, they were packed in layers in specimen holders, and enclosed in SS 316L tubes. Then, they were inserted in the SINQ target (Target-4), within the Pb spallation target rods, as schematized in Fig 1. All samples were irradiated for 16 months, with a total proton charge of 10.03 Ah. Further specifications on the irradiation conditions of the Ta specimens (irradiation experiment "IB") are given in [33]. A list of the STIP-II Ta samples retrieved and available for separation is given in Table 2. In this study, the IB18, IB19, IB22, and IB55 specimens were treated.
Radioactive tracers for γ-spectrometry. The separation process was monitored via γspectrometry. The γ-emitting radionuclides contained in the irradiated Ta targets and their main γ-lines monitored during the separation are reported in Table 3.
Additionally, γ-emitting Lns tracers were added to monitor the separation of Sm, Gd, and Dy, respectively (see Table 4).
For each separation, ca 1-5 kBq of each radiotracer was placed in a HDPE scintillation vial, brought to dryness at 70˚C under N 2 flow, and then re-dissolved in 0.5 mL 0.05 M HCl. γspectroscopic measurements were performed with a high-purity germanium (HPGe) detector. Energy calibration of the spectrometer was executed with a certified 152 Eu standard source from PTB (Physikalisch-Technische Bundesanstalt, Braunschweig, Germany). γ-spectra were analyzed with the Genie 2000 Gamma Analysis software.
Separation setup. The separation set-up consisted of the solution reservoir (i.e., CELL-STAR1 polypropylene tube, capacity 15 ml), a peristaltic pump, a separation column, and the collection vessel (HDPE material, capacity 20 mL). As separation column, single fritted filtration columns (capacity 3 mL, 10-20 μm filter) filled with the resin of interest were used. Bed heights were as follows: LN resin-3 cm; DGA resin-1 cm; SYKAM resin-5 cm; LN3 resin-1 cm.  Table 3. List of γ-emitting radionuclides contained in the irradiated Ta targets. For each radionuclide, the half-life (T 1/2 ), the energy (E γ ) and intensity (I γ ) of the main γ-lines monitored during the separation are indicated. Data taken from [44].

Handling of radioactive material
Separation experiments were performed in the PSI Hotlab, in a lead shielded fume hood, applying the ALARA-As Low As Reasonable Achievable-principle. In each step, the dose rate was kept as 20 uSV/h (measured at the working distance), using dose rate meters 6150AD 1 Automation und Messtechnik GmbH. The personnel was equipped with personal dosimeters (for the detection of gamma, beta, and neutron irradiation), operational dosimeters (electronic, with immediate dose display and with audible indication of the radiation level when thresholds for dose or dose rates are exceeded), as well extremity dosimeters (finger rings integrating a thermoluminescence-based detector).

Separation procedure
The first step in the separation procedure concerned the dissolution of the radioactive Ta specimen, followed by the removal of the Ta matrix. Then, two fractional separation methods were applied, namely Method A and Method B. Method A was conceived as a direct fractional separation of the lanthanides after their extraction from the Ta matrix. However, due to the high activity of the treated samples, an alternative separation procedure was considered (Method B). Further details are given in the following sections. Dissolution and removal of the Ta matrix. The procedure for the dissolution of the Ta samples and the precipitation of the Lns fluorides is described in detail in [43]. The dissolution of the "Tensile" type specimen was carried out in 4 mL 10 M HNO 3 and 4 mL conc. HF. When the same mixture of acids was applied to the "Bend bar" type samples, only a partial dissolution together with fluoride precipitation was obtained. The supernatant was thus removed and transferred to another centrifuge tube. Then, the remaining part of the target was dissolved in additional HNO 3 and HF. To increase the Lns separation yield, precipitation steps were repeated in supernatants by adding Lu carrier. The formation of a LuF 3 precipitate helped the co-precipitation of the other lanthanides as LnF 3 . The obtained precipitate was separated from the supernatant, rinsed and collected. The precipitate was washed with Milli-Q water to remove any remaining Ta. The scheme for the LnF 3 precipitation is shown in Fig 2. LnF 3 was dissolved in a centrifuge tube with 1 mL 0.5 M H 3 BO 3 and 1 mL 6 M HNO 3 . H 3 BO 3 was used to bind any remaining HF and to facilitate the dissolution of the aqueous insoluble lanthanide fluorides by forming tetrafluoroboric acid in solution [45]. To facilitate the dissolution the tube was placed in a water bath at 80˚C. After dissolution, alternatively Method A and Method B were applied in order to sequentially isolate the desired Lns fractions from the LnF 3 precipitate.
Separation Method A. A schematic representation of the separation Method A is given in Fig 3. The pH of the dissolved LnF 3 was adjusted to 1 with 3 M NH 4 OH, and the γ-emitting tracers were added. The solution was then loaded into the Sykam macroporous cation exchange resin. After loading,~12 mL of 1 M NH 4 NO 3 was used to bring the column pH to~5. Table 4. List of γ-emitting radionuclides tracers added to monitor the separation of Sm, Gd, and Dy. For each radionuclide, the half-life (T 1/2 ), the energy (E γ ) and intensity (I γ ) of the main γ-lines monitored during the separation are indicated. Data taken from [44].

ICP-MS measurements
The ICP-MS measurements were performed using an Element II1 Sector-Field Inductively Coupled Plasma Mass Spectrometer (SF-ICP-MS) by Thermo Fisher. The samples were diluted in 0.28M HNO 3 prepared from high purity nitric acid and Milli-Q water. Dilution factors (DF) of 10, 50, or 100 were chosen after pre-measurements (see Table 5).
This step was necessary for having in the measurement solution a suitable amount of the main isotopes to determine their concentration, while still being able to detect other lanthanide impurities by scanning the background. As an internal standard, a certified standard solution with 10 mg/L Re was added (ESI, Elemental Scientific, Omaha, NE, USA). The concentrations of the main isotopes in the solution were determined with an external multielement standard (TraceCERT1, Periodic Table Mix 1, 2, and 3) prepared in different concentrations (dilution series). The ICP-MS settings (gas flows, torch position, and lenses) are tuned daily by introducing an in-house multi-element solution. From this solution, B, Rh, and U (1 ppb each) are used to optimize intensity, peak shape, and U/UO value to minimize oxide  formation (3-4%). The mass offset for each isotope was checked by measuring the multi-element standard solution after tuning the machine. These mass offsets in the range of -0.1 to +0.1 were imported to the operation method. The operation method was set to scan the masses between 116 and 187 several times in low-and medium-resolution. Instrument settings and data acquisition parameters are shown in Table 6.
In the measurement sequence, the samples are bracketed with the measurement of the multi-element standard dilution series. Each sample, as well each standard series, is preceded by a HNO 3 -blank. Each sample was measured twice independently. The detection limit of each isotope is calculated with three times the standard deviation of the Blank measurements and the dilution factor of the samples is considered. The results of the measurement are given as intensities for each mass in counts per second (cps), average of the scans. All count rates are corrected for machine background contributions by subtracting the signals previously measured on pure acid solutions. Internal normalization to cancel drift in responsiveness of the machine is done relative to 185 Re (internal standard). Then, the normalized count rate versus the concentration for each mass of the external multi-element standards series (measured before and after the samples) is interpolated and applied to calculate the concentration (ppb) of each sample. The relative uncertainty of the concentrations is empirically stated as 10% (2σ). The latter includes the variations of standard measurements treated as samples, the weighing uncertainties in the preparation of the standard and sample solutions, as well the uncertainty of the concentrations in the external standard (� 0.8%, 2σ).

Results and discussion
Due to the long time elapsed after irradiation of the Ta samples (i.e., 17 years) short-lived reaction products were already decayed. Exemplificative γ-spectra of the LnF 3 precipitate and the supernatant after fluoride precipitation are shown in Fig 5. As in [43], an average 98% Lns separation yield from the Ta matrix was here obtained. The detection of 133 Ba and 178m2 Hf radionuclides in the LnF 3 precipitate (Fig 5A) is due to the sorption of Hf and Ba cations on the negatively charged surface of the LuF 3 carrier, that ultimately leads to the co-precipitation of Hf and Ba in the form of fluoride complexes [46]. The LnF 3 precipitate showed a contact dose rate between 0.5 and 30 mSv/h, depending on the type of Ta target initially dissolved. This dose-rate mainly derived from the decay of 172 Hf (and its daughter 172 Lu), 178m2 Hf, 173 Lu, and 133 Ba.
In Method A, the dissolved LnF 3 was directly loaded into the Sykam resin. Prior to the loading, the pH adjustment of the sample was performed since the distribution coefficients (K d ) of Lns in cation-exchange resins increase with lowering the acidity of the loading solution. At such conditions, the competition for the resin binding sites between the lanthanide ions and the hydroniums of the loading solution is minimized, and thus, all the Lns are retained in the resin phase. After passage of 1 M NH 4 NO 3 and~10 mL 0.09 M HIBA, the pH of the column was adjusted to~5. Only at this point, the elution of the Lns was observed. This is due to the fact that at strong acidic conditions the -COOH functional group of HIBA exists predominantly in un-dissociated form, and thus the ligand shows a poor coordination strength [47]. The extraction of the Lns from the resin was performed by complexation with HIBA. An example of the separation profile with Sykam resin is presented in Fig 6. The removal of the Lns from the Sykam resin was performed by applying a gradient elution technique. With increasing the HIBA concentration, the number of molecules in the mobile phase capable of chelating the Lns (III) ions is augmented. This results in a lower affinity of the Lns ions for the cation exchanger, and hence, in their elution from the separation column. Because of the lack of Nd, Pr, Ce, and La γ-emitter tracers, these elements could not be individually separated and hence were eluted together with Pm by using 0.25 M HIBA, and collected in the "Pm fraction". Satisfactory peak separations were obtained at an elution flow rate of 0.4 mL/min. At higher flow rates, overlap of peaks due to longitudinal diffusion were observed. After performing the fractional separation, each lanthanide of interest was eluted from the Sykam resin within~20 mL of HIBA. After the Lns separation, Ba was stripped from the column with 4 M HNO 3 . The concentration of each lanthanide in a reduced volume (~5 mL) of 1 M HNO 3 was performed with the LN3 resin column. The upload in HIBA phase exploited the high affinity of the resin towards Lns(III) ions in low concentrated acids [48], whereas 1 M HNO 3 was sufficient to elute the Lns.
In the separation Method B, the extraction of Hf, Lu, and Ba from the fluoride precipitate was prioritized in order to decrease the dose rate to which the personnel was exposed. For the removal of Hf and Lu, LN resin was chosen due to the high K d of Hf(IV) and Lu(III) on HDEHP resin [49,50]. An example of γ-spectra showing the qualitative separation of Hf and Lu with LN resin is shown in Fig 7. With this separation step, a reduction of~80% of the initial emitted dose was achieved. After elution from the LN resin, the Ba and the Lns fraction (4 M HNO 3 ) was directly loaded onto the DGA resin. It is known that Lns(III) show a high affinity for the DGA resin, regardless of the used HNO 3 concentration [51]. At the same conditions, Ba(II) is instead not retained. In order to elute the Lns from the resin, a different mobile phase is thus needed. It was previously shown that Lns heavier than Nd have a log 10 (K d ) > 3 on DGA resin for HCl concentrations above 4 M [51]. However, when the concentration of HCl is lowered, the affinity of Lns towards the DGA resin is dramatically reduced. Thus, the elution of the Lns was pursued with 0.05 M HCl. The collected Lns were directly loaded to the Sykam column, previous addition of the γ-tracers. Due to the low acidity of the 0.05 M HCl solution, its pH adjustment was not necessary. After loading the Lns in the Sykam column, passage of~35-40 mL of 0.09 M HIBA was in average required to adjust the pH of the column to~5. The Lns fractional separation followed the same procedure as previously described in Method A, with the difference that no Hf and Lu elution peaks were observed (Fig 8).
The purity of the collected Dy, Tb, Gd, and Sm fractions was analyzed by ICP-MS. Comparative ICP-MS results are shown in Figs 9-12. Only signals above the limit of detection (LOD) were considered.
Chemical purity of each fraction was checked by assessing the absence/presence of the nuclides reported in Table 7.
These nuclides were selected as benchmarks since they are not affected by isobaric interference, i.e., each atomic mass (A) corresponds only to a single nuclide.
In all the fractions, the disruption of the natural isotopic abundance of each element, with a predominance of neutron-deficient isotopes, was observed. ICP-MS analysis showed a significant presence of Lu (~100 times the LOD) in the Dy and Tb fractions obtained with Method A. The high amounts of Lu contained in the LnF 3 precipitate led to the tailing of the Lu elution peak, which affected the quality of the Dy and Tb fractions. Presence of Hf nuclides (signals at   Dy at A = 158 Dalton are expected. By increasing the Sykam column bed height, a better separation resolution might be achieved when an initial Hf, Lu, and Ba separation from the LnF 3 precipitate is not applied. However, this would comport a considerable increase of the time required for the separation, and hence, an unnecessary exposure of the personnel to high dose rates.
In the Sm, Gd, and Dy fractions obtained with Method B, no isobaric interferences are detected. In the Tb fraction, however, a signal 3 times above the LOD at A = 161 Dalton-i.e., 161 Dy-was recorded. Due to the detection of Dy in the sample, the presence of the 158 Dy isotope cannot be excluded. From the isotopic abundances in the Dy fraction, a 161 Dy/ 158 Dy ratio of 1.61 was deduced. Since the amount of 161 Dy in the Tb fraction corresponds to 1.90 ppb, a  Table 8, together with the minimum amount of material needed for half-life determination studies.
The remaining irradiated Ta samples (equivalent overall to 45.29 g of Ta material) will be processed by applying the separation Method B.

Conclusions
A radiochemical separation method for retrieving ppb level of Sm, Gd, Tb, and Dy from Ta targets irradiated with proton and spallation neutrons was developed. It was proven that the removal of Ba, Hf, and Lu at the beginning of the process was necessary not only to minimize the overall dose rate exposure (in the mSv/h range), but also to obtain pure Dy and Tb fractions (i.e., devoid of Lu and Hf contaminants). ICP-MS analysis showed a disruption of the natural isotopic abundances, with a predominance of neutron-deficient nuclides such as 146 Sm, 148,150 Gd, 154 Dy, 157,158 Tb. No isobaric interference was determined in the purified Sm, Gd and Dy fractions. In the Tb fraction, 1% of Dy (mass fraction) was detected. However, the contribution of 158 Dy to the 158 Tb-158 Dy isobaric interference was reported as less than 5%. The purified lanthanide fractions are suitable for the preparation of samples for nuclear physics experiments, i.e., for the preparation of α-or γ-sources for half-life measurements. During the separation process, fractions containing Ba, Hf, Lu, Ho, Eu, and Pm radionuclides were collected as well. Further separation of these fractions is envisaged.   Table 7. Masses of reference (in Dalton), corresponding to nuclides not affected by isobaric interference. The identification of the reported mass signals in the ICP-MS spectra corresponds to the univocal presence of a specific element and thus, to the related stable and long-lived isotopes.

Mass(es) of reference (Dalton)
Assigned