Gas-phase molybdenum-99 separation from uranium dioxide by fluoride volatility using nitrogen trifluoride

Production of the important 99mTc medical isotope parent, molybdenum-99 (99Mo), via the fissioning of high- and low-enriched uranium (HEU/LEU) targets followed by target dissolution in acid and solution-phase purification of 99Mo is time-consuming, generates quantities of corrosive radioactive waste, and can result in the release of an array of radionuclides to the atmosphere. An alternative 99Mo purification method has been devised that has the potential to alleviate many of these issues. Herein, we demonstrate the feasibility of a rapid Mo/Tc gas-phase separation from UO2. The results indicate that volatile [99Mo]Mo can be captured downstream of the reacted solid mixture on a column bed (trap) of alumina; the majority of the captured [99Mo]Mo can be subsequently eluted from the alumina trap with a few milliliters of water. >1.0 × 105 single pass decontamination of U and the collected [99Mo]Mo product is demonstrated. This simple thermo-fluorination technique has the potential to provide a rapid methodology for routine 99Mo production.


Introduction
Technetium-99m ( 99m Tc, t 1/2 ¼ 6.01 h) is the most widely used diagnostic radionuclide world-wide. It is dispensed at radiopharmacies and hospitals via 99 Mo/ 99m Tc generators. 1 The 99 Mo (t 1/2 ¼ 65.98 h, $6.1% ssion yield) parent is produced via the ssioning of highly enriched uranium (HEU) targets. The 99 Mo is chemically puried from the target material and other ssion and activation products by processing of the acid-dissolved HEU targets. 2 Processing of the dissolved HEU targets requires several days, produces corrosive liquid waste streams, and an enriched uranium waste stream. Further issues related to chemical processing for planned conversions of HEU target materials to LEU targets have been discussed by Vandergri and other researchers. 2,3 In this article, we discuss the volatility proles and separation protocol for the 99 Mo/ 99m Tc couple from uranium using nitrogen triuoride. In most of the proposed uorination methods such as the FLUOREX process, 4 or those described earlier by workers in the Czech Republic, 5 the former Soviet Union, 6 or the US, 7-9 uorine gas is used to rapidly form volatile UF 6 from an irradiated matrix, generally UO 2 or U metal 10 A large set of volatile uoride products such as PuF 6 , NpF 6 , IF x (for x ¼ 3, 5, 7), TeF 6 , TcF 6 , etc., can be binned cryogenically, or can be sorbed onto solid traps that have specic capture affinities for the volatile products. 10,11 Regardless of the exact process used, uorination requires the use of rigorously closed reactor and trapping systems. These systems are thus more suited to complete trapping of volatile ssion products than the liquid digestion processes currently in use.
The usefulness of NF 3 for volatility separations is related to its slightly lower thermal reactivity compared to more potent uorinating reagents. The lower reactivity of NF 3 allows for volatility of a reduced set of ssion products, in particular, Mo and Tc, without volatilization of U, Np and Pu. The basis for the separations is the formation of thermally stable, nonvolatile UO 2 F 2 produced as the rst product in the uorination of UO 2 (eqn (1)), or of UF 4 in the case of the U metal uorination (eqn (2)): The uorinated solid matrix formed per eqn (1) or (2) can be further reacted with NF 3 to extract volatile uorides without formation of gaseous UF 6 . The onset temperature for the conversion of UO 2 F 2 to UF 6 is usually near 500 C, but can be stalled nearly completely by lowering the NF 3 concentration to 1 or 5% in Ar. 12 Similarly, in a metal target the conversion of UF 6 from UF 4 can be considerably slowed using reduced temperature or lower NF 3 concentrations. 13 This feature of the reactions allows for gaseous leaching of the solid U-bearing sample for the time required to volatilize lower boiling point components (such as 99 Mo/ 99m Tc) that are generally shown to be rapidly separated at or below 400 C. Separation and recovery of volatile MoF 6 /TcF 6 from other ssion products has been demonstrated by use of selective sorbents, such as solid MgF 2 . 14 High temperature oxidation of irradiated U has been widely cited as being effective at removal of gaseous ssion products such as Xe and Kr. 15 This is more rapidly and completely realized by the lattice disruption of the U solid, as induced by uorination (eqn (2) and (3)). Fluorination using NF 3 will volatilize Nb, Mo, Sb, Tc, Te 16,17 and I from a solid matrix at or below $400 C. Ru will be released near 500 C. 18 Rhodium, Pd 18 and Pu 17 do not form volatile uorides using NF 3 as the uorination reagent. Americium, 19 the lanthanides, and the Group I and II elements do not form volatile uorides using any uorinating reagent. 20 Aer down-stream capture of the volatile ssion products has been performed, uranium can be recovered as gaseous UF 6 per eqn (3) or (4), leaving the lanthanides, Pu, Am, and the other non-volatiles in the reactor furnace.
Here, we show that a gas/solid leaching process using NF 3 to recover 99 Mo/ 99m Tc from a simulated UO 2 target has a sound empirical basis that promises rapid, single pass, high yield recovery of 99 Mo/ 99m Tc. began to form quickly in the presence of the reducing agent. Aer several hours, it was determined that the Mo(VI) / Mo(IV) reduction was complete. Tube 1 was centrifuged at $8000 rpm using Sorvall MC 12V centrifuge (Dupont, Newtown, CT). Next, the supernate was removed. Water (1 mL) was added to tube 1, and the [ 99 Mo] MoO 2 crystals were re-suspended. Depleted uranium dioxide powder (23.05 mg) was added to the tube and the [ 99 Mo]MoO 2 / UO 2 mixture was thoroughly mixed by sonication. Tube 1 was again centrifuged and the supernate discarded. The mixture was re-suspended in $250 mL water, and 50 mL aliquots of the suspension were added to a Ni sample pan that had been placed under an infrared heat lamp. The solid suspension was quantitatively added to the pan in successive $50 mL aliquots as the liquid in the pan was evaporated. Once thoroughly dried, the Ni pan containing the mixture of [ 99 Mo]MoO 2 and UO 2 was transferred to a thermo-uorination apparatus for gas-phase [ 99 Mo]Mo separation from UO 2 .

Thermo-uorination apparatus
Thermogravimetric (TG) and differential thermal (DTA) screening data for the reaction of NF 3 on samples of UO 2 , Mo and Tc metal, MoO 2 and TcO 2 , and MoO 3 (Fig. 1B) was acquired using a modied Seiko TG/DTA 320. 12 The gases used for thermoanalytical experiments were 99.995% purity NF 3 from Advanced Specialty Gases (Reno, NV) and 99.9995% ultra-high purity (UHP) Ar from OXARC (Pasco, WA). The same instrument was used in the reduction of 99 TcO 2 to 99 Tc metal, wherein a stream of 4% H 2 (99.99%) (OXARC) in Ar was used at 600 C for 1 h.
Modication of the TG/DTA system included conditions for adequate gas mixing and improvements for corrosion resistance. NF 3 /UHP Ar gas mixtures were premixed in 4 linear feet of SS tubing (0.25 inch OD) prior to their entry into the furnace chamber of the TG apparatus. The premixed gas was routed through the analytical microbalance chamber by a 1/16 inch OD nickel tube to an area about 2.54 cm from the sample and reference pans. This distance reduced buoyancy motion of the sample and reference arms as the dense gas mixture was released from the nickel tube and also allowed for some laminar ow of the gas mixture along the direction of the sample. A larger UHP Ar ow was passed though the analytical balance and sensitive electronic components to protect them from a backow of hot NF 3 and other reaction product gases. Three ow meters were used to adjust the NF 3 /Ar concentration to a total gas ow rate of 200 mL min À1 . The platinum thermocouples inside the balance beams were plated with nickel and the plating was covered in ceramic. The coatings help to reduce hot NF 3 corrosion of the thermocouples for extended reaction screening of Mo, Tc and U samples, below 550 C. The coatings were supplied by RT Instruments (Woodland, CA). were contained in the tubing with the use of Monel screens placed within the tube unions; the spheres were slightly crushed to produce trap packing media that allowed unimpeded ow of gases. Behind the rear trap, a quartz wool plug was placed aer the rear Monel screen. From there, a Teon tube routed effluent gases through a 125 mL Erlenmeyer ask congured as a water bubbler.
At the end of a thermo-uorination experiment, the traps and the furnace tube were disassembled, and each component was washed using a series of solvent washes as is described in detail below. The residual components in the nickel sample pan were fully analyzed by dissolution of the entire sample pan in nitric acid. The distribution of 99 Mo was evaluated by gamma counting, and that of the U was evaluated by inductively coupled plasma-mass spectrometry (ICP-MS).

Radiometric measurements
1. HPGe: reference standards were prepared by spiking known volumes of 99 Mo-bearing solutions (in secular equilibrium with 99m Tc) into 2.0 mL of 0.1 M HCl in 20 mL glass scintillation vials. These samples were analyzed using several high purity germanium (HPGe) gamma detectors (Ortec, Oak Ridge, TN) that had been energy and efficiency calibrated for this geometry  using NIST traceable standards. Gamma spectra were evaluated using Genie 2000 Gamma Acquisition and Analysis soware (v. 3.4.1) (Canberra, Meriden, CT). The mean 99 Mo activity obtained in the reference standards using the HPGe detector analysis was used to establish the various detection efficiencies (E d ) for 99 Mo-bearing samples of non-standardized geometries using NaI(Tl) scintillation detectors (described below). 2. Auto-gamma counter: aqueous samples were prepared as 2.0 mL aliquots in 12 Â 74 mm test tubes for counting on a Wizard 1470 (PerkinElmer, Meriden, CT) automatic gamma counter containing a well-type NaI(Tl) scintillation detector. The detector was congured with a counting protocol specic to the 99m Tc gamma emission region of interest (corresponding to 140.57 keV (89 AE 4% intensity)). Samples were not analyzed until secular equilibrium between 99 Mo and 99m Tc was attained (sample analyses were performed $24 h aer each experiment was conducted). The E d for the Wizard 1470 was determined by comparing the count rate of a 2.0 mL aliquot of 99 Mo/ 99m Tc solution in the test tube vs. the 2.0 mL aliquot activity determined by the calibrated HPGe detector.
3. Benchtop NaI(Tl) detector/scaler: non-aqueous samples (e.g., sample pan, trap components, furnace tube) were counted using a Ludlum 2200 scaler/rate meter coupled to a 2 00 dia. NaI(Tl) scintillation detector (Sweetwater, TX). Sample observation distance was maximized to $15 cm to minimize geometry effects. At a given sample/detector distance, E d was determined by comparing the NaI(Tl) detector count rate with that of the HPGe-analyzed standard as described above. Again, samples were not analyzed until secular equilibrium between 99 Mo and 99m Tc was attained.

Mass spectrometric measurements
Aer complete decay of 99 Mo, dilutions of the dissolved Ni sample pan and trap leachates were prepared in 2% Optima grade HNO 3 . Quantication of U in the diluted solutions was performed by an Agilent 7700X (Ventura, CA) ICP-MS. Sample solutions were delivered to the mass spectrometer with a uoride-resistant polyuoroalkoxy alkane sample intake and nebulizer (Glass Expansion, Pocasset, MA). A ten-point calibration curve was prepared by gravimetric dilutions from a NIST traceable 1000 ppm single element U standard obtained from   3 Thermo-fluorination conversion of UO 2 to UO 2 F 2 to UF 6(g) with 5% NF 3 . UO 2 F 2 to UF 6(g) conversion occurs at temperatures well above that required for volatile MoF 6 / 99 TcF 6 formation. High Purity Standards (Charleston, SC). The calibration curve had a regression coefficient of 0.9999.

Evaluation of Mo, Tc, and UO 2 volatility by uorination
Fluoride volatility of Tc was likewise evaluated under the same conditions. Fig. 2A shows the evolution of 99 TcF 6 from 99 Tc metal, which initiates at $180 C, and that from 99 TcO 2 , which initiates at or below $250 C isothermal in Fig. 2B. The volatile reaction products are analogous to the Mo complexes discussed above. In Fig. 1B and 2A and B, the volatile species of Mo and Tc were removed from the reaction system by the Ar gas purge as demonstrated by the steep downward slopes of the TG scans. The Mo and Tc volatility proles were found to be quite similar, and the complete removal of Te, Ru, Nb, Sb, and several other elements have been shown previously to follow suit. 17,18 The behavior of UO 2 with exposure to NF 3 provides a stark contrast to that observed with Mo and Tc species, as shown in Fig. 3. Fluorination of UO 2 is quite unique to this oxide of U and has been described previously by members of this research team. 12 Using the same 5% NF 3 /Ar mixture employed for Moand Tc-bearing materials, UO 2 was converted to non-volatile UO 2 F 2 once the temperature approached 420 C, aer which a plateau region was sustained for several hours with the proper NF 3 exposure conditions before signicant production of gaseous UF 6 occurred. The thermogravimetric evaluations with gas streams of heated 5% NF 3 /Ar indicate that gas-phase separations of Mo (metal and MoO 2 ) and Tc (metal and TcO 2 ) from UO 2 is feasible.
Gaseous uorides of these transition metals can be generated at temperatures below the conversion temperature of UO 2 to UF 6 (via UO 2 F 2 formation). This permits NF 3 leaching of a ssioned UO 2 solid with no UF 6 attendant in the gaseous Mo (Tc) phase.

Gas-phase separation of [ 99 Mo]MoO 2 from UO 2
Given the preceding thermogravimetric results for metal and metal oxide constituents and UO 2 , a gas-phase separation of 99 Mo (as MoO 2 ) from UO 2 was evaluated. A sample was prepared in a nickel sample pan that consisted of a homogeneous mixture of ne UO 2 (23 mg) and MoO 2 crystals (1.5 mg); NaBH 4 was initially used to reduce an aqueous solution of Na 2 (5) For this experiment, the outlet of the modied TG furnace tube was connected to tandem traps (A and B) that were packed with activated alumina. A third trap (C) was packed with a compact bundle of quartz wool (Fig. 1A). A uorination experiment was performed with a 5% NF 3 /Ar mixture, and the furnace temperature held at $400 C for 2 h. At the end of the experiment, the trapping system components were disconnected from the furnace tube, and each of the three traps was disconnected from each other. Next, each component in Fig. 1A was leached using a series of solvent washes. These washes included that of the furnace tube and each of the three traps. The nickel sample pan (and salt residues) was completely dissolved in nitric acid. The water in the bubbler trap was acidied and evaporated to near dryness. Each component and wash solution was analyzed by gamma counting ( 99 Mo/ 99m Tc) and ICP-MS (U).
Analysis of the distribution of [ 99 Mo]Mo and U revealed an excellent separation of the ssion product from the simulated ssioned source material. The Mo was almost completely removed from the sample pan, with only 4% remaining ( Table  2). Approximately 10% was deposited on the walls of the furnace tube, and 71% was captured in Trap A. Less than 1% of Mo was measured in Traps B, C, and the bubbler. In total, 86% of the Mo was accounted for in the assays of the trapping components. Of the Mo captured in Trap A, $70% was removed with a 5 mL H 2 O rinse (representing $50% of the total Mo pan deposit), and an additional 21% was recovered in two sequential washes with NaOH (Table 3). Within the three Trap A aqueous washes, $65% of the pan-deposited Mo was recovered.
Radiometric counting of the trapping components immediately aer disassembly (before 99 Mo/ 99m Tc secular equilibrium was attained) provided qualitative indication that 99m Tc was transported efficiently out of the pan and was successfully deposited primarily in the furnace tube and Trap A. Unfortunately, quantitative determination of the 99m Tc depositions were not possible with the use of the NaI(Tl) scintillation detector/scaler. However, an HPGe detector scan of the post-reacted Ni sample pan indicated that 99m Tc was successfully volatilized and transported out of the pan, thereby corroborating the observed volatilization prole shown in Fig. 2. In sharp contrast, the U remained in a non-volatile state; 95.3 AE 3.4% of the original U deposit remained in the nickel sample pan, and $0.027% was found in the furnace tube (0.024%) and the three traps (0.003%, Table 2). Based on the mass of U measured in the combined trap leaches, the U decontamination factor in the Trap A [ 99 Mo]Mo product was >1.0 Â 10 5 . Total U recovery in all fractions was found to be 95.5 AE 3.2%, a value that was within the analytical uncertainty of the experiment.

Conclusions
We show that exposure of a homogeneous mixture of [ 99 Mo]Mo/ UO 2 to 5% NF 3 /Ar mixture at $400 C for $2 h results in a rapid, high yield extraction of [ 99 Mo]Mo from U.
Of the $86% of 99 Mo activity accounted for in the various furnace/trap components, $71% of the 99 Mo activity was deposited in the rst alumina trap. A simple 5 mL water wash of the trap's alumina bed resulted in $70% of the trapped 99 Mo activity removal, which represented $50% of the total 99 Mo activity originally deposited in the nickel pan. Technetium-99m was likewise transported and collected on the alumina trap with the separated 99 Mo product, although quantitative distribution was not possible in this rst test. The results indicate that the gas-phase [ 99 Mo]Mo product was largely devoid of U contamination.
Aqueous processing releases I, Te, Xe and Kr potentially at every step of processing of irradiated targets. Acid dissolution, in particular promotes, volatile behavior in several elements as Tc, and Ru. While uoride volatility must release these species as well, we believe that the front-end processing of irradiated uranium targets by volatility-based separations is better suited by its rigorous closed engineering to sequester radionuclide populations than the digest and back end, clean-up approach historically and currently used by most nuclear-related enterprises. Fluoride volatility separations of 99 Mo from uranium, so described, has a sound chemical basis. Its practical implementation for radiopharmaceutical scale processing still requires elucidation of transport and capture technologies that are optimized for high efficiency retention of isotopes of pharmaceutical interest.

Conflicts of interest
There are no conicts to declare.