Large-Scale Production of 119mTe and 119Sb for Radiopharmaceutical Applications.

Radionuclides find widespread use in medical technologies for treating and diagnosing disease. Among successful and emerging radiotherapeutics, 119Sb has unique potential in targeted therapeutic applications for low-energy electron-emitting isotopes. Unfortunately, developing 119Sb-based drugs has been slow in comparison to other radionuclides, primarily due to limited accessibility. Herein is a production method that overcomes this challenge and expands the available time for large-scale distribution and use. Our approach exploits high flux and fluence from high-energy proton sources to produce longer lived 119mTe. This parent isotope slowly decays to 119Sb, which in turn provides access to 119Sb for longer time periods (in comparison to direct 119Sb production routes). We contribute the target design, irradiation conditions, and a rapid procedure for isolating the 119mTe/119Sb pair. To guide process development and to understand why the procedure was successful, we characterized the Te/Sb separation using Te and Sb K-edge X-ray absorption spectroscopy. The procedure provides low-volume aqueous solutions that have high 119mTe-and consequently 119Sb-specific activity in a chemically pure form. This procedure has been demonstrated at large-scale (production-sized, Ci quantities), and the product has potential to meet stringent Food and Drug Administration requirements for a 119mTe/119Sb active pharmaceutical ingredient.

Introduction

Recent efforts in using radioactive isotopes in vivo have provided creative solutions to numerous global health problems.[1−18] Consider that positron and X-ray emissions from isotopes like 18F, 82Rb, 68Ga, 99mTc, and 201Tl now find widespread use in imaging technologies to treat millions of patients worldwide each year.[19−22] Equally exciting is the potential for harnessing particles emitted during nuclear decay to treat disease, e.g., cancer, bacterial infections, viral infections (like HIV), and other nonmalignant disorders (such as degenerative skeletal pain, Graves orbitopathy, and Gorham Stout syndrome).[23,24] Of numerous radionuclides that show promise, 119Sb is particularly interesting. This isotope decays by emitting K-edge and conversion electrons, collectively called Auger electrons. The 119Sb attraction originates from the low energy (∼20 keV) of these Auger electrons, which results in short biological path lengths (∼10 μm) that are comparable with the diameter of a many human cells.[25] Hence, therapeutic targeting with 119Sb provides a unique opportunity to deliver a lethal dose of radiation to a targeted diseased cell while leaving the adjacent healthy tissue unharmed.[26−30] The potential for patient recovery along with little to no hematological toxicity (no negative side-effects) is extraordinary in comparison to nontargeted treatment methods, i.e., nontargeted chemotherapy.

One of the most pragmatic challenges facing implementation of 119Sb in medical applications is associated with access. Today 119Sb can be produced at certain cyclotron facilities in reasonable quantities (0.1–1 Ci).[31,32] Production routes typically involve irradiation of isotopically enriched tin-119 (119Sn) targets (eq ). Unfortunately, the brisk (relatively short) 119Sb half-life [38.19(22) h][33] and the somewhat complicated and lengthy 119Sn/119Sb separation limit the time interval over which usable activity is available for distribution (Figure ).Identifying alternative methods that prolong access to 119Sb would expand distribution to medical institutions that do not have colocated 119Sb production facilities. The impact could be dramatic, and transition 119Sb drug development from a niche area of research into a medical therapeutic akin to commercially available Azedra[34] and Xofigo,[35] which use 131I and 223Ra as active agents.

Figure 1

Plot showing how 119Sb isolated from the 119mTe parent (red and green traces) generated at a high-energy proton source prolongs access time to 119Sb directly produced from 119Sn (blue trace) at a common cyclotron.

Recent nuclear cross-section measurements suggest alternative 119Sb production routes exist that could prolong access to 119Sb.[36] These predictions maintain 119Sb could be made in large quantities (10–100 times larger than the cyclotron-based routes described above) through the nuclear reactions described in eqs –4 using high-energy proton sources, i.e., the Isotope Production Facility (IPF) at the Los Alamos Neutron Science Center (LANSCE) at Los Alamos National Laboratory (LANL) and the Brookhaven Linac Isotope Producer (BLIP) at Brookhaven National Laboratory. The proposed approach involves addition of a proton to the two naturally occurring isotopes of Sb, namely, 121Sb and 123Sb. Subsequent neutron loss generates 119mTe, three in the case of 121Sb and five for 123Sb. Removal of the natSb target material leaves behind 119mTe [t1/2 = 4.70(4) days],[33] which slowly decays to the 119Sb daughter by electron capture, eq .

Using natSb targets to produce 119mTe (and subsequently 119Sb) has several advantages over low-energy irradiations of 119Sn targets that form 119Sb directly (eq ). The natSb targets do not require costly isotopic enrichment; the higher proton energies needed for their irradiation at facilities like IPF and BLIP offer proton fluxes and integrated fluences that are orders of magnitude larger than those used at small cyclotron facilities, and a generator system is possible that uses 119mTe to make 119Sb available for weeks from a single production process. Figure demonstrates this graphically by comparing 119Sb activities generated from two conservative (on the low side of what is possible) production campaigns. One production effort involves cyclotronic irradiation of 119Sn to directly produce 0.1 Ci of 119Sb. The other indirectly generates larger quantities of 119Sb (1 Ci) through the 119mTe decay. In these scenarios, small quantities (0.001 Ci) of 119Sb are available from the 119Sn route after ∼10 days. In contrast, 119Sb (0.001 Ci) lingers after ∼45 days when prepared using the 119mTe/119Sb approach, more than 4 times longer than the cyclotron method. Even if 119Sn production of 119Sb generated equivalent quantities of 119Sb to the 119mTe routes, the 119mTe approach is still more attractive because the longer 119mTe decay rate prolongs access to the 119Sb daughter, eq . The combination of “holding-back” 119Sb by way of its 119mTe parent and increasing production yields with the (p,3n) and (p,5n) reactions (eqs and 3) offers a technical solution to the “access” problem that plagues 119Sb drug development.

Herein, we experimentally confirm propositions that the 119mTe and 119Sb pair can be generated from natSb in large quantities by describing a production and separation method that can greatly increase availability. In this manuscript the 119mTe/119Sb production process is described in its entirety, from target design to final product isolation. For example, informed by previous excitation function measurements,[36−40] we established irradiation conditions and an appropriate target that was compatible with large-scale 119mTe production (Ci-sized quantities). On the separation side, our initial efforts were unsuccessful. They either failed to accommodate the large (25 g) Sb target or were incompatible with the remote handling techniques required to safety manipulate the highly radioactive product. To overcome these challenges, we adopted an unconventional approach in isotope production and made use of X-ray absorption spectroscopy (XAS) to guide efforts that removed Te from Sb. Spectroscopic results informed researchers on differences between Te and Sb in aqueous solutions matrixes and on solid-state chromatographic supports. The information was used to develop a Te/Sb separation procedure that (1) was compatible with large natSb (25 g) targets, (2) accounted for safety concerns associated with generating large quantities of 119mTe/119Sb, and (3) could be rapidly (∼36 h separation) carried out using remote handling techniques within hot cells.[41] We are most excited by the fact that the final 119mTe/119Sb product could be isolated as a chemically pure sample, in reasonable yield, and with high specific activity. Provided an appropriate 119Sb generator can be designed that meets strict requirements from the Food and Drug Administration (FDA), this 119mTe/119Sb production route also shows potential for compatibility with active pharmaceutical ingredients (APIs). We hope these results will provide a basis for expanding 119Sb medical applications and provide researchers insight that supports isotope production and separations for other rare isotopes that show promise in medical applications.

Results and Discussion

Research and Development Approach

In developing a procedure for separating 119mTe from natSb we adopted the unconventional experimental approach of uniting isotope production science with XAS spectroscopy. The goal was to identify experimental conditions under which Te and Sb existed in chemical forms that facilitated separation using chromatographic methods compatible with large-scale processing (25 g Sb, 2.7 Ci or 5.9 × 1016 atoms of 119mTe) and remote handling techniques, i.e., within hot cells. To this end, we used X-ray absorption near edge structure (XANES) to assess oxidation states and extended X-ray absorption fine structure (EXAFS) to determine local coordination environments for macroscopic amounts (>1 mg) of Te and Sb. This spectroscopic duo revealed differences in charge and chemical forms for Te vs Sb and identified reactions accessible to Te that were inaccessible to Sb. A separation procedure was subsequently developed based on the unique chemical reactivity we observed for Te. To communicate these results most effectively, we first present XANES and EXAFS studies on the aqueous behavior of Te and Sb in solutions that showed promise for the separation (HCl and HF). Next, we describe the developed production method. We conclude by reporting on Te reactivity with three chromatographic resins (CL, Rare Earth, and AG 1-X8) used in the procedure.

Te and Sb K-Edge XAS Spectroscopy in HCl and HF

The Sb oxidation state was characterized as a function of HF concentration. First, a stock solution was obtained by dissolving Sb0 (metal shot) in a mixture of HNO3 (16 M) and HF (28 M). Second, HNO3 was removed and the solution media converted to concentrated HF (28 M). More dilute HF solutions were prepared from this 28 M stock. Oxidation states were determined using Sb K-edge XANES spectroscopy by comparing the inflection point energies from the analytes with oxidation state reference standards, namely, SbV2O5, SbIII2O3, and SbIIICl3 (Figures and 3, and see the SI). This approach enabled SbV to be distinguished from SbIII, as the inflection points for the +5 and +3 oxidation states differed by 3–4 eV (see the SI).[42,43] All spectra obtained from Sb dissolved in HF solutions exhibited inflection points close to 30 495 eV, suggesting that the predominant Sb oxidation state was +5, independent of HF concentration (Figures and 3, and see the SI). Converting the solution matrix to HCl did not appreciably alter the Sb inflection point energies. The highest-energy inflection point observed was in 2 M HCl at 30 496.8(7) eV. Increasing the Cl1– concentration to 12 M systematically decreased the inflection point energy to 30 493.8(7) eV. Although the inflection point energies for Sb in concentrated HCl solutions were ∼1 eV lower in energy than SbV2O5, they were still ∼2 eV higher in energy than SbIIICl3 and SbIII2O3 standards, suggesting that the dominant Sb oxidation state in HCl solutions was also +5.

Figure 2

Dependence of the inflection point energies on acid concentration from solution phase Sb (top) and Te (bottom) K-edge XANES measurements in HF (red) and HCl (blue) solutions. Measurements from Sb and Te standards and Te that has been loaded onto various resins (anion, Rare Earth, and CL) are shown in black. Uncertainties were estimated at 0.2 eV for Te and 0.7 eV for Sb based on our ability to reproduce these values.

Figure 3

Normalized Sb (left) and Te (right) K-edge XANES from aqueous solutions (black traces; top—HCl, 0–12 M; bottom—HF, 0.01–28 M) as well as for Te fixed to the CL (red trace), anion (blue trace), and Rare Earth (green trace) resins. The SbCl3, Sb2O5, Te metal, TeO2, and Te(OH)6 oxidation state references are shown as gray traces.

To characterize chemical identities for primary species present in HCl solutions, shell-by-shell fitting analyses of Sb K-edge EXAFS spectra were performed (Figure ). When the Cl1– concentration was highest (HCl = 10–12 M), the EXAFS spectra were dominated by a single frequency. This data was best modeled by a shell of 5.8(4) Cl atoms with a 2.370(5) Å Sb–Cl distance (Table ). Our measured bond distances were shorter than those observed in K3SbIIICl6 (ca. 2.52 Å)[44] and SbIIICl21+ [2.42(2) Å],[45] which was consistent with the SbV oxidation state assignments derived above.[42,43] Hence, we concluded SbVCl61– was the dominant species present in concentrated HCl matrixes (10–12 M).

Figure 4

Sb and Te K-edge EXAFS (a) and Fourier transform–EXAFS (b) from aqueous solutions (HCl, 0–12 M) as well as from Te fixed to the anion and Rare Earth resins. Experimental spectra are shown in black and fits in red. (c) Graphical representation showing the O (blue ●), Cl (red ●), and the total (○) coordination numbers dependence on HCl concentrations for Sb (left) and Te (right).

Table 1

Fitting Parameters for Sb in Solution with 2–12 M HCl

	Sb–Cl
sample	S02	CN	R (Å)	σ2 (Å2)
2 M HCl	0.95a	1.0 ± 0.4	2.37 ± 0.018	0.0033b
4 M HCl	0.95	2.4 ± 0.3	2.39 ± 0.014	0.0033
6 M HCl	0.95	3.7 ± 0.2	2.38 ± 0.008	0.0033
8 M HCl	0.95	4.8 ± 0.5	2.37 ± 0.008	0.0034 ± 0.0010
10 M HCl	0.95	5.7 ± 0.3	2.37 ± 0.005	0.0040 ± 0.0006
12 M HCl	0.95 ± 0.06	5.8 ± 0.4	2.37 ± 0.005	0.0033 ± 0.0006

The amplitude reduction factor was set to 0.95 based on previously reported values (0.95 ± 0.05) for Sb2O5.[46]

The Sb–Cl Debye–Waller factor was constrained based on fits to the 12 M HCl EXAFS spectrum.

The Sb–O Debye–Waller factor was constrained based on fits to the 2 M HCl EXAFS spectrum.

Systematically decreasing the Cl1– content—by decreasing HCl concentration—from 8 to 2 M caused a lower-frequency signal to emerge in the Sb EXAFS spectra. This low-frequency contribution was easily observed in the k-space plots of Figure . The two oscillations at 6 and 8 Å–1 in the 12 M HCl spectrum moved closer together as the Cl1– content decreased until they finally coalesced at 2 M HCl. These low-frequency contributions in k-space gave rise to a peak in the corresponding Fourier transforms at R + Rδ ∼ 1.2 Å. We fitted this peak using a shell of O atoms. The intensity of this new peak steadily increased with decreasing HCl concentration (from 8 to 2 M), such that the O coordination numbers increased from 0.6(4) to 4.8(1.1) while the Cl coordination numbers dropped from 4.8(5) to 1.0(4).

There are multiple ways to interpret these results. One option involved a single SbV compound that contained a combination of Cl1–, OH1–, and H2O ligands. For example, in 2 M HCl, Sb could exist as Sb(OH)4(H2O)Cl. Alternatively, a mixture of two SbV endmember species—SbCl61– in 12 M HCl and Sb(OH)5(H2O) in dilute HCl—could be present in 2–8 M HCl. In the latter scenario, the SbV speciation in 2 M HCl would contain 15% SbCl61– and 85% Sb(OH)5(H2O). Increasing the HCl concentration would then systematically increase the percentage of SbCl61– and decrease the percentage of Sb(OH)5(H2O), (Figure ). Additional confidence in our interpretation comes from congruency with previous reports showing our proposed dilute endmember Sb(OH)5(H2O) exists in aqueous solutions at pH 1.[46] Because EXAFS spectroscopy provides a snapshot of the average Sb coordination environment across all species present in a sample, we could not use this data to distinguish between these two scenarios nor exclude formation of intermediates between the SbCl61– and Sb(OH)5(H2O) endmembers.[47]

To determine Te oxidation states in relevant aqueous media, a stock solution containing natural-tellurium (natTe) was prepared for XAS analyses. Samples were made by dissolving Te0 (metal shot) in a mixture of HF (28 M) and HNO3 (16 M) followed by HNO3 removal. This stock was used to prepare aqueous natTe samples with varied HF (0.01–28 M) and HCl (2–12 M) concentrations. Subsequently, Te oxidation states were determined using Te K-edge XANES spectroscopy, as the inflection point energies for TeIV and TeVI typically differ by ∼4 eV (Figures and 3, and the SI).[47−49] The 31 817.0(2) eV inflection point measured for natTe in 0.01 M HF was only 0.6 eV higher in energy than TeIVO2 and substantially lower (by almost 4 eV) in energy than the TeVI(OH)6 standard. Increasing the HF concentration slightly increased the inflection point values by ca. 0.5 eV. Over the entire HF concentration range, the inflection point values averaged 31 817.5 with a standard deviation of only 0.1 eV (1σ). In HCl, the inflection point energies were lower (by ca. 1.5 eV) relative to Te in straight H2O. These HCl values remained relatively constant over the entire HCl concentration range, averaging 31 815.5(4) eV between 2 and 12 M. Hence, this data suggested TeIV was the predominant species present in all HF (0.01–28 M) and HCl (2–12 M) solutions.

As with Sb, the Te K-edge EXAFS spectra from samples in HCl (0–12 M) were modeled. The Te EXAFS spectra were fitted with inner sphere O and Cl atoms, and the number of O and Cl atoms coordinated to Te varied as a function of Cl1– concentration. At high HCl concentrations (8, 10, 12 M) only Cl atoms were detected: 4.0(4)–4.6(5) Cl atoms with Te–Cl distances ranging from 2.46(1) to 2.49(1) Å. These distances were consistent with expectation based on ionic radii, i.e., summing the 0.66 four-coordinate TeIV radius with the 1.81 Å Cl1– radius equaled 2.50 Å. Furthermore, our results agreed reasonably well with previous EXAFS measurements made on TeIV in solutions with high Cl concentrations (12 M LiCl and 1 M HCl),[47] which showed 4.3(4) inner sphere Cl1– with a 2.504(5) Å Te–Cl bond distance. Owing to measurement uncertainties (1σ), Cl1– coordination numbers were bound between 3.6 and 5.1, and we could not rule out the presence of TeCl51– and TeCl62–. Prior UV–vis and Raman spectroscopy studies suggested that the TeCl62– species dominated at high HCl concentrations.[50,51]

In contrast, in dilute HCl (0–2 M) no inner sphere Cl1– ligands were detected; only O atoms were observed. The coordination numbers [3.6(3) and 3.2(2)] and Te–O distances [1.89(1) and 1.90(1) Å] were consistent with tellurous acid (H3TeO31+) being the major species. This interpretation agreed with previous work by Grundler and co-workers. These authors examined Te speciation in 9.9 M HClO4 solutions (pH < 0) using EXAFS spectroscopy. They found that H3TeO31+ was the dominant species with a Te–O bond distance of 1.90(1) Å.[49]

At intermediate HCl concentrations (4 and 6 M), the Te K-edge EXAFS results showed Te coordinated by both O and Cl atoms. The number of coordinated Cl1– ligands increased (and the number of O atoms decreased) with increasing HCl concentration (Table ). Results from the fitting analyses could indicate that a single Te species existed whose O and Cl coordination numbers systematically varied with HCl concentration (total coordination number of ∼4, see Figure and Table ). Alternatively, a mixture of TeCl4 and H3TeO31+ could exist in solution, and the ratio of TeCl4 to H3TeO31+ would change as a function of HCl content. Another interpretation acknowledges intermediate compounds exist between the high (TeCl4) and the low (H3TeO31+) HCl concentration extremes, like TeCl4OH1– and TeCl2(OH)2. Again, the composition of the mixture would vary with HCl concentration. The latter interpretation was consistent with Raman and UV–vis data reported previously and suggested mixtures of Te species were present in solutions at moderate HCl concentrations (2.5–8.5 M), including TeCl4OH1– and TeCl2(OH)2.[51]

Table 2

Fitting Parameters for Te in Solution with 0–12 M HCl

	Te–Cl
sample	S02	CN	R (Å)	σ2 (Å2)
0 M HCl	0.9a
2 M HCl	0.9
4 M HCl	0.9	2.3 ± 0.5	2.49 ± 0.02	0.0058 ± 0.0025
6 M HCl	0.9	3.7 ± 0.5	2.49 ± 0.01	0.0067 ± 0.0017
8 M HCl	0.9	4.0 ± 0.4	2.49 ± 0.010	0.0075 ± 0.0013
10 M HCl	0.9	4.6 ± 0.5	2.48 ± 0.011	0.0097 ± 0.0014
12 M HCl	0.9	4.5 ± 0.4	2.46 ± 0.011	0.0098 ± 0.0013
Rare Earth	0.9	4.8 ± 0.3	2.51 ± 0.005	0.0074 ± 0.0007
Anion	0.9	4.3 ± 0.3	2.51 ± 0.006	0.0050 ± 0.0008

Amplitude reduction factor set based on ref (47).

Debye–Waller factor set based on fits of EXAFS spectrum for Te in 2 M HCl.

Results from the XAS analyses provided the basis for developing separation strategies for removing 119mTe from natSb (Scheme ). Consider, at low HCl concentrations Te existed primarily as a cationic complex whereas Sb was primarily a neutral compound. At high HCl concentrations a neutral complex dominated the Te speciation, and an anionic complex dominated for Sb. We also surmised that at intermediate HCl concentrations the major Te species in solution involved cationic and neutral compounds, while Sb complexation involved neutral and anionic species. These charge differences provided the opportunity to separate Te from Sb based on ion-exchange chromatography and liquid/liquid extraction schemes.

Scheme 1

Diagram Showing the Average Sb and Te Stoichiometries for Species Present between 0 and 12 M HCl

Te and Sb Separations

An natSb metal target (25 g; 1.52 in. diameter; 0.136 in. thick) was sealed in an Inconel capsule (window thickness = 0.012 in. per side) designed to withstand high-energy irradiations at the Isotope Production Facility (IPF) at Los Alamos National Laboratory (see the SI). The target was irradiated (19.51 h) with an average proton beam current of 100 mA (total charge ca. 1885.4 mA h). Prior to irradiation, the optimum proton beam energy was determined by irradiating a monitor foil positioned at the target entrance position, the 100 MeV incident beam was then degraded to an average value of 42.5 MeV using precision machined aluminum metal spacers. After irradiation, the target was transferred to a hot cell. Next, the irradiated natSb (natural isotopic distribution) material was removed from the target shell, and the natSb metal was dissolved in a mixture of nitric (HNO3, 15.7 M) and hydrofluoric (HF, 28 M) acid for subsequent processing.

Using insight from the Sb and Te K-edge XAS data acquired on aqueous solutions (described above) and solid-state chromatographic resins (described below), the following Te/Sb separation method was developed (Scheme ). The solution matrix was converted to pure HF in preparation for passage through a column loaded with the commercially available CL resin (Eichrom Technologies). Although this matrix change represented the most time-consuming step in the procedure—accounting for 10 h of the ∼36 h process—it was essential for debulking Sb from the 119mTe analyte. If the solution contained residual HNO3 (from the target dissolution) the separation would fail, and 119Te would leach through the resin. However, when a pure HF (28 M) solution was loaded onto the CL resin bed, a blue band formed at the top of the column; 119mTe was retained by the resin, and natSb eluted (Figure ). Under these conditions, macroscopic quantities of the natSb (demonstrated at 50 g with a mock target) target could be almost completely removed (more than 99.994%) from microscopic amounts of 119mTe (2.69 Ci, 11.5 μg). After the initial separation on CL resin, recovering 119mTe was not straightforward. It could not be eluted off the column in HF, HCl, or even water. However, in HNO3 (≥10 M) the 119mTe isotope eluted in near quantitative yield (99%, see the SI). Although this approach successfully accommodated production-scale quantities of natSb and 119mTe, it was initially concerning that the CL resin introduced ppm quantities of S and P into the sample. Small-scale experiments used for developing the remote handling methods and the large-scale procedure showed (by ICP-AES) that the 119mTe eluate contained S and P contamination near 600 and 200 ppm, respectively. Fortunately, we overcame this issue by adding additional S and P removal steps, vide infra.

Scheme 2

Overall Separation Scheme for 119mTe from an natSb Target

Figure 5

Chromatograms from the Te/Sb separations utilizing CL, Rare Earth, and anion-exchange resins. (left) Separation of 121Te (3240 cps) from stable Sb (160 mg) traced with 124Sb (114.5 cps) using a Bio-Rad column (10 mL column) containing CL resin (Eichrom Technologies; 1.5 mL; 100–150 μm). (middle) Separation of stable Sb (12 mg), 124Sb (32 cps), and 121mTe (260 cps) using a Bio-Rad column (10 mL column) containing Rare Earth resin (Eichrom Technologies; 1.5 mL; 50–100 μm). (right) Separation of stable Sb (12 mg), 124Sb (32 cps), and 121mTe (260 cps) using a Bio-Rad column (10 mL) containing AG 1-X8 anion-exchange resin (Bio-Rad; 1.7 mL; 100–200 mesh).

After eluting 119mTe from the CL resin, the solution matrix was rapidly changed to concentrated HCl (11.7 M) for a second Te/Sb separation. This involved loading the sample onto a Rare Earth resin (Eichrom Technologies) in concentrated HCl. Under these conditions 119mTe was retained, and natSb unexpectedly fractionated (Figure ). Using 124Sb radiotracers to characterize the Sb elution profile in developmental experiments, we observed approximately half of the 124Sb was retained on the column, while the rest passed through the column (Figure ). This natSb fractionation was of little consequence in terms of 119mTe recovery because 119mTe could be eluted cleanly from the Rare Earth resin in H2O, in high yield, and without natSb contamination. As testament, during developmental experiments using a combination of natSb (160 mg) and 124Sb, Sb decontamination factors were unable to be determined as there was neither 124Sb nor natSb detected in the 119mTe elution fractions by γ spectroscopy or ICP-AES. The 119mTe yield in these developmental experiments was 97%.

To ensure complete separation of 119mTe from the natSb target, we found it prudent to include an anion-exchange-based (AG-1 × 8 resin, Bio-Rad) separation step.[52] Developmental experiments showed this anion-exchange step provided an additional Sb decontamination factor (initial Sb concentration ingoing ÷ final Sb concentration in product) of over 900 (Figure ). The anion-exchange separation was also attractive because it was fast (5 h), high-yielding (119mTe recovery at 97%), and easily incorporated into the large-scale procedure. For example, in the large-scale process, the 119mTe eluate from the Rare Earth column was acidified with HCl (to 2 M). To ensure high-oxidation states were retained for the analytes in this load solution, peroxide (H2O2) was added and the solution gently heated. The solution was then loaded onto the anion-exchange resin. Under these conditions natSb passed directly through the column; 119mTe was retained, and subsequently eluted with H2O.

The final step in the 119mTe purification was designed to remove S and P contamination using Prefilter resin (Eichrom Technologies). We observed previously that this resin can significantly decrease organic contaminants in samples.[53,54] Thus, the eluate from the anion-exchange column was loaded directly onto Prefilter resin and 119mTe eluted. Experiments used for development demonstrated that S and P were removed completely from solutions containing 0.003 and 0.005 mg of S and P, respectively. In these experiments the S and P levels in the eluate were below our limit of detection (∼100 ppb for both elements by ICP-AES).

Te K-Edge XAS Spectroscopy from Columns

To understand the origin for the chromatographic results described above, a series of Te K-edge XAS measurements were conducted on the first three columns deployed in the separations shown in Scheme . In each experiment, macroscopic quantities of naturally occurring tellurium (natTe; ca. 1 mg) were loaded on a resin that was contained in a small column. The entire column was placed in an aluminum holder fixed to a N2(l) cryostat and the XAS spectra collected at low temperature. Given that macroscopic quantities of natTe behaved similarly to microscopic amounts of 119Te on the three resins we have confidence in correlating spectroscopic results from natTe (macroscopic) with the 119Te (microscopic) separations.

Analysis of natTe on a CL resin column (used in Step 1 of Scheme ) provided surprising results. The extractant adsorbed on the CL resin (R3P=S) was not used to directly bind TeIV. Instead, the Te K-edge XANES spectrum indicated that Te was retained by the resin as Te0 metal (Figure ). The inflection point from the resin-retained natTe [31 813.6(3) eV] was nearly identical to that for metallic Te0 at 31 814 eV. These values were ca. 4 eV lower than TeIV in the HF (28 M) load solution (vide supra).[55] The spectrum’s line shape was also consistent with the spectral features of metallic Te0, in that the absorption peak was substantially attenuated and resembled a simple step function. This interpretation was confirmed by fitting the Te K-edge EXAFS spectrum as metallic Te0 and by subsequent powder X-ray diffraction measurements, which showed diffraction peaks consistent with formation of Te0 on the resin.[56] To the best of our knowledge this represents the first observation of CL resin achieving separations through electron transfer reactions—as opposed to simple extraction chemistry—and we are currently exploring new potential application of this material.

Similar to the Te K-edge EXAFS results from natTe bound to CL resin, examination of the natTe retained on the Rare Earth resin (Step 2 of Scheme ) also provided unexpected results. The EXAFS spectrum revealed a single peak, located at approximately the same distance as the peaks in the EXAFS spectra from the 10 and 12 M HCl samples. These similarities suggested that Te was retained by the Rare Earth resin as an anionic chloride complex and showed no evidence of the Rare Earth resin using its octyl(phenyl)-N,N-diisobutylcarbamoyl-methylphosphine oxide (CMPO) extractant to form molecular complexes with TeIV, like a Te–CMPO compound. Fitting the Rare Earth resin spectrum revealed TeIV coordinated by 4.8(3) Cl atoms at a distance of 2.510(6) Å. This coordination number was slightly higher than that determined for TeIV in 8 M HCl solutions, 4.0(4), and was equivalent within acceptable error to those values determined for the 10 and 12 M HCl solutions (Figure ). Additionally, the Te–Cl bond distance was slightly longer (by 0.02 Å) for TeIV retained on the Rare Earth resin than TeIV dissolved in 6, 8, 10, and 12 M HCl solutions, which also corresponded well with the increased Cl coordination number. Overall, these results showed the Rare Earth resin acted as a simple anion exchanger under our experimental conditions, and the expected extraction chemistry was not observed.

Interactions between natTe and the anion-exchange resin (AG 1-X8, Bio-Rad) (Step 3, Scheme ) were also characterized by Te K-edge EXAFS after loading natTe in 2 M HCl onto the resin. The spectroscopic results described above suggested that, when dissolved in 2 M HCl, Te would exist predominantly as a cationic species (H3TeO31+). Thus, retention of Te by the anion-exchange resin was perplexing because the resin should only bind anions. Analysis of the anion-exchange resin-retained Te provided insight into the reason for the chromatographic behavior. The dominant peak in the Fourier transformed EXAFS spectrum from retained Te was fitted with Cl atoms and the small shoulder on the main peak fitted with O atoms. The O coordination number was 0.8(1), and the Cl coordination number was 4.3(3). Although the Te–O distance for the sorbed species was equivalent to the HCl solution phase measurements, the Te–Cl bond distance (2.510(6) Å) was slightly longer than the Te–Cl distances in HCl solutions, suggesting that the Te–Cl bond lengthened to accommodate a greater number of ligands in the Te coordination sphere (similar to our observations of the Rare-Earth-retained Te). These results suggested that species with an average stoichiometry of TeCl4OH1– were retained on the resin. Furthermore, the results indicated that contacting TeIV in 2 M HCl with an anion-exchange resin causes a change in speciation, from H3TeO31+ to TeCl4OH1–. Motivated by this observation, we are currently characterizing the complicated reaction chemistry at the anion-exchange resin–HCl mobile phase interface.

Outlook

The 119mTe processing method reported here is compatible with performing routine, large-scale production of 119mTe and 119Sb (through the decay of the parent isotope 119mTe). The procedure appropriately accounted for safety concerns associated with the significant radiation doses accompanying large-scale accelerator produced 119mTe and 119Sb, in that it could be carried out using remote handling techniques in a hot cell. Despite these handling challenges, this procedure also enabled 119mTe and 119Sb to be isolated rapidly (∼36 h from target dissolution to final product suspension), which is important given the brisk half-lives for 119mTe [4.70(4) days][33] and 119Sb [38.19(22) h].[33] The final product was obtained in small volumes of aqueous solution (14 mL), with high specific activities (119mTe activity/total mass Te), and in chemically pure forms. No stable Te was detected in the final product by ICP-AES (the minimum detectable concentration in final product was 100 ppb), and the only contaminants observed were phosphorus (5 ppm) and natSb (2 ppm).

To aid in developing the 119mTe/119Sb production method, significant effort was expended on understanding the separation process. These studies revealed unexpected insight into key separation steps that were responsible for successful 119mTe processing (Scheme ). Consider our analyses on the CL resin (Step 1, Scheme ) used to debulk 99.994% of the 25 g natSb target material from the 119mTe product. We assumed initially that the CL resin achieved 119mTe separation through standard methods, i.e., forming R3P=S—Te bond between 119mTe and the adsorbed R3P=S extractant. However, the Te K-edge XAS results and powder X-ray diffraction studies showed this was not the case. Instead our results revealed a new reaction pathway available to the CL resin, one that involved an electron transfer reaction to reduce TeIV to Te0. These results demonstrate that CL resin can achieve separations as advertised (by extraction) or unconventionally (observed herein) by acting as a nonconducting stationary support capable of redox chemistry. The impact of this newly observed separation pathway for the CL resin extends beyond processing of 119mTe, potentially opening doors for additionally innovative solutions to complicated partitioning problems.

Insight into CL resin redox chemistry also explained numerous peculiarities observed during the Te/Sb separation. Prior to the XAS studies, S and P contamination from the CL column was mysterious as was the origin of a blue band that formed on the top of the column. We were additionally confused by observations that HCl, HF, and H2O were ineffective at freeing Te from the CL resin bed because TeIV is quite soluble in these matrixes. Reduction of TeIV to metallic Te0 provided a likely explanation for all of these observations. Redox reactions between the R3P=S extractant on the CL resin and TeIV could introduce P and S decomposition products into the column effluent. Other redox induced decomposition products likely accounted for the blue color on top of the resin. Finally, reduction of TeIV to metallic Te0 explained the unusual elution profile. Metallic Te0 will only elute if dissolved. Contacting the resin with HNO3 (10 M) achieves dissolution, whereas HCl, HF, and H2O do not.

The XAS analysis of retained Te on the Rare Earth (Step 2, Scheme ) and anion-exchange (Step 3, Scheme ) resins provided additional insight into the Te/Sb separation procedure. For instance, like the CL resin, we anticipated that the Rare Earth resin would use the adsorbed CMPO extractant to bind TeIV. However, this was not observed. Instead, Te K-edge EXAFS spectroscopy showed that TeCl51– was retained rather than a TeIV–CMPO coordination complex. In hindsight, this observation seemed reasonable as the CMPO extractant was likely protonated in 12 M HCl, leaving the resin to act as a simple anionic ion exchanger.

The AG 1-X8 anion-exchange column provided the only example where a resin deployed in our 119mTe/119Sb process behaved as expected, at least for the most part. For instance, although anionic exchange partitioning was observed, the results were unexpected based on the aqueous speciation summarized in Scheme . The EXAFS analysis of the Te/Sb AG 1-X8 separation was thought-provoking regarding resin performance. We remind the reader that (1) TeIV and SbV were loaded onto the AG 1-X8 anion-exchange resin in 2 M HCl, and (2) Te K-edge EXAFS analysis showed that Te existed primary as H3TeO31+ in 2 M HCl. This cationic species, having no Cl1– and 3.2 ± 0.2 oxygen ligands, should not have been retained by an anion-exchange resin. Indeed, it was not. Instead, the TeIV speciation changed from H3TeO31+ in the mobile phase to an anionic TeIV species on the resin, with an average stoichiometry of TeCl4OH1–. This observation contributes to the growing body of knowledge describing complex—and difficult to characterize—chemistry that occurs at interfaces, e.g., between solids and solutions or between two immiscible liquids.[57−61] We propose two possible pathways that rationalize the TeIV behavior. In the first scenario, the alkyl ammonium (NR41+) functional group on this solid-state support may force the Cl1– concentration at the surface of the resin bead higher than in the 2 M HCl mobile phase. The increased Cl1– concentration could change the identity of the dominant TeIV species, increase the number of Cl1– ligands bound to TeIV, and favor formation of the electrostatically bound TeCl4OH1–. The second scenario acknowledges that EXAFS spectroscopy results on the 2 M HCl mobile phase show an average stoichiometry of TeCl4OH1–. Other species may be present in small quantities, like TeCl4OH1–, TeCl51–, and TeCl62–. It seems possible that the positively charged NR41+ functionalized resin may selectively bind small quantities of TeCl4OH1– (or TeCl51–/TeCl62–) present in 2 M HCl, thereby shifting the TeIV speciation at the resin/solution interface. Both scenarios are likely oversimplified. However, these hypotheses were useful in that they provided frameworks to rationalize the retention of TeIV on the anion-exchange resin using the EXAFS data, which unambiguously demonstrates that dominant TeIV species present in the mobile phase were not the major species bound to the solid-state support. We cannot overstate the importance of better characterizing solid–solution interfaces present within separation processes, as the chemical compositions at these assemblages are critical for controlling analyte partitioning. The impact spans beyond that of fundamental curiosity, ranging from fate and transport of toxic ions in the environment, environmental restoration, and (in this case) isotope production for medical purposes.

In summary, the new separation capability for 119mTe and 119Sb represents an exciting leap forward for targeted radiotherapeutic use of Auger-emitting isotopes. It offers opportunity to greatly increase access to 119Sb and expand 119Sb availability beyond the relatively limited number of medical institutions that have 119Sb production capabilities. The potential impact of these results is significant, offering opportunity to further explore 119Sb targeted Auger therapies, to identify appropriate methods for administration (either chelated or nonchelated), and to transition 119Sb Auger therapy from a niche area of R&D into mainstream medical procedures used to treat disease. The EXAFS results for CL resin performance are also exiting, particularly the revelation that redox reactions can be used to separate Te from Sb. Hence, we are currently trying to expand Te-based electron transfer reactivity to solve challenges in 119mTe/119Sb generator design. The anticipated results have potential to provide a reliable 119Sb/119mTe generator and further enhance 119Sb access to researchers and clinicians.