Identification of a chemical fingerprint linking the undeclared 2017 release of 106Ru to advanced nuclear fuel reprocessing.

The undeclared release and subsequent detection of ruthenium-106 (106Ru) across Europe from late September to early October of 2017 prompted an international effort to ascertain the circumstances of the event. While dispersion modeling, corroborated by ground deposition measurements, has narrowed possible locations of origin, there has been a lack of direct empirical evidence to address the nature of the release. This is due to the absence of radiological and chemical signatures in the sample matrices, considering that such signatures encode the history and circumstances of the radioactive contaminant. In limiting cases such as this, we herein introduce the use of selected chemical transformations to elucidate the chemical nature of a radioactive contaminant as part of a nuclear forensic investigation. Using established ruthenium polypyridyl chemistry, we have shown that a small percentage (1.2 ± 0.4%) of the radioactive 106Ru contaminant exists in a polychlorinated Ru(III) form, partly or entirely as β-106RuCl3, while 20% is both insoluble and chemically inert, consistent with the occurrence of RuO2, the thermodynamic endpoint of the volatile RuO4 Together, these findings present a clear signature for nuclear fuel reprocessing activity, specifically the reductive trapping of the volatile and highly reactive RuO4, as the origin of the release. Considering that the previously established 103Ru:106Ru ratio indicates that the spent fuel was unusually young with respect to typical reprocessing protocol, it is likely that this exothermic trapping process proved to be a tipping point for an already turbulent mixture, leading to an abrupt and uncontrolled release.

In the fall of 2017, the man-made, high-yield fission product 106Ru (half-life, t1/2 = 373.6 d) was detected by monitoring networks across Europe (1–4), along with sporadic detections of minute amounts of the relatively short-lived 103Ru (t1/2 = 39.2 d) in select locations. Although unprecedented in scale (250 TBq) (5), airborne and surface measurements substantiated a timely assessment that there was no detrimental impact to human health (6). Nevertheless, the undeclared intrusion of such radioactivity into the air space and soil of sovereign nations demands investigation, in support of national and coordinated global security.

To address the location of the 106Ru source, recent reports have used dispersion modeling and field measurements of 106Ru concentration (airborne and ground deposition) to demonstrate an origin in the Southern Urals of Russia, in the area of the Mayak industrial complex (1, 5–9). A long history of nuclear-related activities in this area, combined with the radiopurity of the field observations and the detection of the short-lived 103Ru, lends credence to the scenario of an accidental release during nuclear fuel reprocessing and serves to dispel some persistent theories on the origin of the release (e.g., nuclear reactor accident, downed radioisotope thermoelectric generator satellite, volatilized medical sources, etc.) (1). However, to make a direct link to fuel reprocessing activity on this basis alone is circumstantial and as such provides room for plausible deniability. Direct evidence to this end constitutes the identification of unique signatures. From a radiological perspective, there is none. Samples have been shown to be radiopure and to carry the stable ruthenium isotopic signature of civilian spent nuclear fuel (10), while stable elemental analysis by scanning electron microscopy and neutron activation has revealed no detectable anomalies compared to aerosol filter media sampled prior to the advent of the 106Ru contaminant (1, 11). We are, then, left with the definition of a limiting case for a nuclear forensic investigation.

Fortunately, we are concerned with an element that has significant covalent character to its bonding interactions (12). This affords the opportunity to perform chemical transformations, with and without the presence of a stable form of the same element. By selecting reactions that are well understood and/or by varying reaction conditions, we can compare the distribution of the radioactive element to the stable form in the reaction product(s) (Scheme 1). In this way, it is possible to deduce both general and highly specific information about the atomic connectivity, that is, the chemical context, of the radioactive contaminant, given that the rules for reactivity are well understood.

Scheme 1.

High-level concept depicting the reactive incorporation of both stable (green) and radioactive (red) chemical species of an element into isostructural end products.

Results and Discussion

To help refine synthetic targets amenable to the strategy proposed in Scheme 1, we subscribe to the hypothesis that the release of 106Ru into the environment is associated with fuel reprocessing activity. Owing to its complex chemistry and high specific activity (typically 6 to 9% of fission products) in spent nuclear fuel (12–16), the removal of radio-ruthenium from waste streams poses a significant challenge. While a myriad of approaches has been proposed, the vast majority center about the PUREX (plutonium–uranium extraction) process where the recovery of uranium and plutonium is accomplished by liquid extraction with an organophosphate from fuel that is digested in nitric acid (16–18). To amplify the oxidative power of the mixture, oxidants such as ceric ammonium nitrate can be added to encourage the evolution of the volatile and highly reactive RuO4, which can then be driven off with carrier gas and subsequently trapped or passivated. At this early dissolution stage, and later during the vitrification of raffinates, the condensation and decomposition of RuO4 to relatively inert oxides inside the containment vessels can lead to problems of accumulated dose and accelerated corrosion (14, 15, 19, 20). To address this, considerable investigative effort has been made to remove ruthenium at the fuel dissolution phase (17, 21, 22). Nevertheless, the formation of RuO4 is still a very appealing option with respect to simplicity, cost, and scalability, as evidenced by recent investigations into ruthenium volatilization through RuO4 generation (23–26). Subsequent chemical reduction of RuO4 to more passive forms is also appealing with respect to such considerations, despite introducing additional waste material for management. Some of the more notable options in this regard are the efficient formation of RuO4−2 and RuO4− salts under alkaline conditions (21, 27), the formation of RuO2 on contact with reducing media (26, 28), and the generation of Ru(III) and Ru(IV) polychlorinated complexes in hydrochloric acid (29–31).

Treatment of filter pieces from both German and Swedish radionuclide surveillance networks with solvent gave highly reproducible results for partitioning of the 106Ru contaminant. Good partitioning reproducibility, in conjunction with autoradiographic imaging (), supports a homogenously distributed contaminant. We have found that 51.3 ± 2.4% of the 106Ru contaminant was associated with the aqueous extract. This result is in very good agreement with that determined by other laboratories (1) and clearly indicates that the 106Ru is composed of more than one chemical form. Partitioning experiments similarly performed with carbon tetrachloride showed no discernable association with the 106Ru contaminant. This indicates an absence of adsorbed RuO4, which is typically soluble in carbon tetrachloride, one of the few solvents that does not succumb to attack by this powerful oxidizing agent (13). This is not surprising, since RuO4 decomposes to RuO2 (hydrated) in water under ambient conditions and further upon exposure to light (15, 32), while reoxidation of RuO2 back to RuO4 is very slight and reserved for atmospheric oxidants such as ozone (32). Interestingly, we found that a small but highly reproducible fraction (7.35 ± 0.70%) of the 106Ru contaminant was partitioned into ethanol. This result further supports a multicomponent composition for the 106Ru contaminant. In light of the aforementioned discussion regarding the pragmatic treatment of RuO4 in nuclear fuel reprocessing, β-RuCl3 is expected to be formed from the passivation of RuO4 in HCl to an extent that depends directly upon time and temperature (30, 31). Among inorganic ruthenium compositions, it is noteworthy that β-RuCl3 possesses the rare quality of being highly soluble in dative solvents. Thus, we have chosen to explore the hypothesis that the passivation of RuO4 in HCl was implicated in the environmental release.

This hypothesis informs an appropriate selection of chemical reactions and synthetic targets toward revealing a unique signature, and other meaningful chemical information, about the 106Ru contaminant. To this end, polypyridyl chemistry is appealing as a reactive vehicle to incorporate 106Ru, since such ruthenium complexes are exceptionally robust and formed in high yield under ambient conditions. Also, characteristic charge-transfer electronic transitions that lie in the visible region of the spectrum make these complexes intensely colored, facilitating chromatographic separation (33, 34). Of the polypyridyl ligands available, those derived from the tridentate 2,2′:6′,2″-terpyridine ligand are particularly well-suited as they offer achiral products and superior kinetic stability (33). Considering the ease of synthesis afforded by one-pot procedures to form 4′-substituted analogs of 2,2′:6′,2″-terpyridine, we have elected to synthesize 4′-p-tolyl-2,2′:6′,2″-terpyridine (ttpy; Fig. 1) according to reported procedures (35, 36), albeit with a notable exception regarding its purification (). With this ligand in hand, we are supported by a plethora of established ruthenium coordination chemistry, the vast majority of which is derived from reaction with β-RuCl3. In particular, we have identified the subsequent monoligated complex, ttpyRuCl3 (37), and the reduced, bis-ligated complex, Ru(ttpy)22+ (38), as targeted reaction products for isolation and radiometric measurement, in adherence to the general strategy presented in Scheme 1. Although the former has invariably served as a reaction intermediate to the formation of heteroleptic complexes (39, 40), it takes center stage in this work. This is due to the fact that the formation of ttpyRuCl3 (or related complexes based upon 2,2′:6′,2″-terpyridine) is expected to be highly selective to β-RuCl3 and related compounds, considering that there is no alteration in oxidation state or local coordination geometry about the ruthenium atom. In fact, to the best of our knowledge, the formation of ttpyRuCl3 has only ever been performed from β-RuCl3. Therefore, by investigating the formation of ttpy106RuCl3 from the relatively small proportion of the 106Ru contaminant that is extractable in ethanol and by demonstrating the highly selective nature of the reaction, we can speak to the existence of β-106RuCl3. In this way, we may determine whether or not passivation of RuO4 in hydrochloric acid was invoked.

Fig. 1.

Synthetic targets used to evaluate the reactivity and distribution of 106Ru.

Radiochemistry 1: Synthesis and Purification of ttpyRuCl3 in the Presence of 106Ru.

With the tridentate ligand (ttpy) in hand, reaction of 1.1 equivalents with β-RuCl3 proceeds in high yield (>80%) from ethanolic solution, provided that the reactant concentration is >0.01 M. The overall radiochemical reaction is presented in Scheme 2 (see also ), and experimental details are provided in . Ethanol containing leached 106Ru (74.5 ± 5.3 mBq 106Ru) was used to synthesize ttpyRuCl3 from stable β-RuCl3 in 88% yield (0.12 g theoretical yield). The insoluble precipitate formed was isolated and counted using a high-purity germanium well detector, using the gamma photon emission at 622 keV (9.93% intensity) associated with the short-lived 106Rh progeny as the analytical signal. Relative to the ethanol extract, 28.8 ± 3.5% of the 106Ru activity was found to be localized in the initial precipitated material. The remaining ethanolic filtrate was chromatographed on silica gel using an eluent mixture of acetonitrile and saturated, aqueous potassium nitrate solution (7:1, respectively) to isolate the sole by-product, Ru(ttpy)22+, after work-up with NH4PF6. This material was similarly gamma-counted and was found to contain 11.2 ± 2.0% (8.37 ± 1.41 mBq) of the 106Ru activity from the ethanol extract. In contrast to the fate of the stable ruthenium, this is a sizeable proportion of 106Ru. However, this is quite reasonable considering that the 0.1 equivalent excess of ttpy used in the reaction (i.e., 7.5 mg) constitutes an enormous molar excess over 106Ru (4.03 × 1012-fold), and that inorganic ruthenium species other than β-RuCl3 will react to form Ru(ttpy)22+ under these conditions. Again, we see evidence of a complex, multicomponent 106Ru contaminant.

Scheme 2.

Reaction targeting the formation of ttpyRuCl3 in the presence of 106Ru. Ru obtained from ethanol extraction. Two hours at reflux temperature. Purified by reverse-phase HPLC. Adjusted for chemical recovery.

While 28.8 ± 3.5% of the ethanol-extracted 106Ru was found to be isolated with the stable product, ttpyRuCl3, consideration of potential flocculation effects precluded the conclusion that ttpy106RuCl3 was formed to any extent. To address this, a high-performance liquid chromatography (HPLC) method was developed to separate ttpyRuCl3 from Ru(ttpy)22+, and from other environmental contaminants associated with the filter material. This HPLC method was first developed on an analytical scale and later adapted to a semipreparative one, using a C18 stationary phase and a mobile phase consisting of an isocratic mixture of N,N′-dimethylformamide (DMF) and methanol (90:10, respectively), optimized at 1.30 mM of tetrabutylammonium chloride (). Note that method development was confined by the limited solubility of ttpyRuCl3 (appreciably soluble in N,N′-DMF, tolerating no more than 30% methanol) and that effective separation of ttpyRuCl3 and Ru(ttpy)22+ was afforded mainly by the addition of the tetrabutylammonium chloride. The characteristic ligand-to-metal and metal-to-ligand charge-transfer bands that characterize ttpyRuCl3 and Ru(ttpy)22+, respectively, were used to arrive at a detection wavelength of 450 nm in the electronic spectrum.

The entirety of the crude ttpyRuCl3 isolated in Scheme 2 (108 mg, 20.9 ± 2.5 mBq 106Ru) was purified by HPLC (Fig. 2). Fractions collected at set times in the HPLC purification process were combined and concentrated for gamma counting. The very small amount of 106Ru measured in these reaction components required long detection count times ranging from 1 to 2 Ms. For this range of count time, the concomitant span of detection capabilities, as defined by the critical limit (Lc) and detection limit (Ld) at 95% confidence, corresponded to Lc = 3.7 mBq and Ld = 7.5 mBq for 1 Ms and Lc = 2.9 mBq and Ld = 5.9 mBq for 2 Ms. Nevertheless, a very good account of the initial 106Ru in the crude ttpyRuCl3 was provided by the components of the HPLC purification process. The elution peak at retention time (Rt) = 2.96 min, corresponding to ttpyRuCl3, was found to contain 10.2 ± 2.6 mBq of 106Ru (Fig. 3), while the fraction collected from 0 to 2.0 min, representing unretained material from the column hold-up volume, was found to contain 12.7 ± 3.1 mBq of 106Ru (). These quantities represent 48.7 ± 13.2% (ttpyRuCl3) and 60.9 ± 14.8% (unretained) of the 106Ru measured in the crude ttpyRuCl3 before purification. The unretained fraction of 106Ru likely consists of unreactive species removed by flocculation as ttpyRuCl3 precipitated from the reaction solution.

Fig. 2.

HPLC purification of ttpyRuCl3 (Rt = 2.96 min) that precipitated from the radiochemical reaction described in Scheme 2. Monitored at λ = 450 nm.

Fig. 3.

Overlay of gamma spectra, in the analytical region of interest (600 to 640 keV) for 106Ru measured in the reaction components leading to the formation and isolation of ttpyRuCl3 (A) and [106Ru(ttpy)2](PF6)2 (B). Depicted are gamma emission peaks corresponding to 214Bi (609 keV, naturally occurring) and 106Ru (616 and 622 keV). (A) Ethanol extract (red); crude ttpyRuCl3 (blue); purified ttpyRuCl3 (green). (B) Filter piece (red); ethanol extract (blue); isolated fraction colocated with added [Ru(ttpy)2](PF6)2 (green).

Radiochemistry 2: Reaction of 106Ru without Stable Ru Precursor.

To provide additional insight and validation, particularly in light of the relatively large counting uncertainty associated with the gamma measurements of the preceding radiochemical experiment (Scheme 2), an analogous experiment was performed without the use of stable ruthenium precursor. In this instance, the formation of the reduced, bis-ligated complex [106Ru(ttpy)2]2+ was targeted (Scheme 3 and ). The reaction to form [Ru(ttpy)2]2+ from 2.1 equivalents of ttpy in ethanol alone was found to be more broadly applicable than previously believed, proceeding smoothly from both Ru(III) and Ru(IV) precursors containing suitable counter anions, as demonstrated by chemical recoveries obtained with β-RuCl3 (89%), (NH4)2RuCl6 (94%), and Ru(NO)(NO3)3 (90%). Thus, we can use an excess of ttpy to compare the reactive fractions of the 106Ru contaminant (extractable in ethanol) obtained from both carrier and carrier-free approaches (Table 1).

Scheme 3.

Reaction to form [106Ru(ttpy)2](PF6)2. Ru obtained from ethanol extraction. Stable complex added postreaction as a tracer for chromatographic isolation. Ethanol, 90 °C, 16 h.

Table 1.

Allocation of 106Ru in the radiochemical reactions relative to the initial quantity measured in the ethanol extract

Reaction	ttpyRuCl3 (% 106Ru, k = 1)	[Ru(ttpy)2](PF6)2 (% 106Ru, k = 1)	Reactive fraction of 106Ru, %
ttpy + β-RuCl3 + 106Ru	14.0 ± 3.9* (16.0 ± 4.4)†	11.2 ± 2.0	27.2 ± 9.0
ttpy + 106Ru	—	28.6 ± 2.6	28.6 ± 2.6

Isolated ttpyRuCl3 fraction from HPLC (Rt = 2.96 min).

Adjusted for chemical recovery.

Here, a mere 0.20 mg of ttpy amounted to an extraordinary molar excess (1.7 × 1010-fold) relative to 106Ru. Following termination of the reaction, stable [Ru(ttpy)2](PF6)2 was added as a tracer to enable chromatographic purification. Gamma counting of the isolated fraction revealed that 28.6 ± 2.6% (0.137 ± 0.013 Bq) of the 106Ru from the ethanol extract was colocated with the added [Ru(ttpy)2](PF6)2. A spectral overlay of the gamma-counted reaction components is presented in Fig. 3. From Table 1, summing contributions of the isolated components from the β-RuCl3 radiochemical reaction (Scheme 2, ttpyRuCl3 and Ru(ttpy)22+), we find a total reactive fraction of 27.2 ± 9.0%. This result is in excellent agreement with the reactive fraction found for the corresponding carrier-free reaction (Scheme 3). Such an intersection of results strengthens further the identification and contribution of ttpy106RuCl3 and additionally serves to provide a more accurate uncertainty estimate considering the higher activity (i.e., lower counting uncertainty) afforded by the carrier-free reaction (Fig. 3).

Formation of ttpyRuCl3: Reaction Selectivity.

While the formation of ttpyRuCl3 has only ever been performed from β-RuCl3 (37, 40), we undoubtedly expect the reaction to proceed to an appreciable extent from higher-order polychlorinated Ru(III) compounds, given the favorable binding interaction afforded by the tridentate ligand, ttpy, and considering that their solution equilibria in dative solvent will include the formation of RuCl3 (41, 42). Nevertheless, by defining the applicable scope of this reaction, we refine and gain confidence in the types of contaminant ruthenium species that could lead to the observed formation of ttpy106RuCl3. Instead of testing an exhaustive selection, several compounds were chosen that reflect a systematic variation from soluble polychlorinated Ru(III) species, namely, variation with respect to ruthenium valency, exchangeable supporting ligands, and physical format. To this end, we had selected the polychlorinated Ru(IV) salt, (NH4)2RuCl6, the Ru(III) mixture, Ru(NO)(NO3)3 (in nitric acid), and the highly insoluble allotrope to β-RuCl3, α-RuCl3. In nitric acid solution, what is denoted commercially as Ru(NO)(NO3)3 is actually a mixture of ruthenium (III/IV) complexes varying in proportion of coordinated and exchangeable nitrate, nitrite, and water molecules, depending on solution conditions (16, 43); moreover, it is directly representative of the nitric acid-based oxidative mixtures used to dissolve spent fuel early in the PUREX process (16, 17). Subjecting these model compounds to the reaction conditions used to synthesize ttpyRuCl3 (), with the notable exception of using saturated ethanolic and aqueous ethanolic solutions of potassium chloride in the case of Ru(NO)(NO3)3, produced no discernable trace of ttpyRuCl3 (Table 2). These tests demonstrate that the reactive tolerance is limited to polychlorinated ruthenium (III) compositions. Therefore, we can conclude that the isolated ttpy106RuCl3 in these experiments originated from 106Ru filter contaminant that exists as β-106RuCl3 or, more generally, [106RuCln(H2O)6-n]3-n. Considering the sequential fractionation of 106Ru in the described experiments, from filter material to ethanolic extract (7.35 ± 0.70%) to precipitated ttpyRuCl3 (28.8 ± 3.5% with 88% chemical recovery) to HPLC isolated ttpyRuCl3 (48.7 ± 13.2%), we determine that 1.17 ± 0.36% of the filter contaminant exists in this chemical form.

Table 2.

Chemical yields of [Ru(ttpy)2](PF6)2 and ttpyRuCl3 from representative inorganic ruthenium compounds

Reactant	Ru ox. state	Solid-state form	[Ru(ttpy)2](PF6)2, % yield	ttpyRuCl3, % yield
Ru(NO)(NO3)3*	3+	Molecular	92	No reaction†
(NH4)2RuCl6	4+	Molecular	96	No reaction
β-RuCl3	3+	Polymeric	94	88
α-RuCl3	3+	Polymeric	95	No reaction
KRuO4	7+	Molecular	19	No reaction
RuO2	4+	Polymeric	No reaction	No reaction

Mixture in nitric acid. ox., oxidation.

Reacted in presence of large excess of potassium chloride.

Assessment of the Bulk 106Ru Contaminant.

That a small proportion of the 106Ru contaminant is composed of polychlorinated 106Ru(III) species is direct evidence that fuel reprocessing was the origin of the 2017 environmental release. Plausible reprocessing activities that could lead to such compounds are limited to either the reductive trapping of oxidatively generated RuO4 in hydrochloric acid (29–31) or the electrochemical reduction and metallization of uranium in spent fuel from molten alkali chloride mixtures (44–46). Fortunately, the compositions formed from both approaches differ substantially. An interrupted process involving the reductive trapping of RuO4 in hydrochloric acid would reasonably be expected to contain some measure of the inert RuO2, owing to the decomposition of RuO4, along with a mixture of Ru(IV) and Ru(III) chloro complexes, the proportion of which will depend upon temperature, duration, and HCl concentration but should favor heavily the reduction to Ru(IV) (30, 31). Mixed nitrosyl–nitrate–nitrite Ru(III/IV) complexes derived from an oxidative nitric acid slurry used to generate the RuO4 may also become entrained upon release, depending on the system engineered. In the case of the pyroprocessing of spent fuel from molten alkali chloride mixtures, chloro complexes are exclusively formed and, although dependent on melt temperature and duration, center upon the formation of Ru(III) chloro complexes with some disproportionation/decomposition to metallic ruthenium occurring very slowly at temperatures >550 °C (44).

By investigating the reactive nature of the bulk of the 106Ru contaminant, we may reveal additional information to help discern between these scenarios. To do this, we have selected commercially available ruthenium compounds representative of the types encountered in such reprocessing activities (Table 2). These compounds comprise a diverse array, differentiated by ruthenium valency, supporting ligand, and physical format, and we can use them to characterize an appropriate reaction (i.e., establish the rules of reactivity). Then, by applying the reactive conditions to the 106Ru contaminant and tracking its subsequent fate and distribution, we may make reasonable deductions about the bulk chemical composition of the 106Ru contaminant and examine how these align with the two plausible scenarios.

Reaction to form the homoleptic complex [Ru(ttpy)2](PF6)2 (Fig. 1) was used as the reactive vehicle. Forceful reaction conditions by way of elevated temperature (154 °C), extended duration (16 h), and provision of dechlorinating agent (AgNO3) were selected to encourage reaction completion (). Interestingly, to date, such reactions have only been performed using Ru (III) (typically β-RuCl3) or specially prepared Ru (II) halogenated compounds (47, 48). Only one literature reference was found pertaining to solvent reduction of a Ru (IV) compound, and this was carried out under microwave irradiation, unrelated to polypyridyl complexation (49). It is therefore quite interesting that this reaction system was found to be very effective for both Ru(III) and Ru(IV) precursors, to the extent that a potent oxidizing agent used in organic synthesis, KRuO4 (50, 51), reacted appreciably while the highly refractory allotrope of RuCl3, α-RuCl3, was found to react smoothly and in essentially quantitative yield (Table 2). Such marked general reactivity using these conditions underscores the complete absence of reactivity observed in the case of RuO2.

Radiochemical details are provided in . For all reactions, we have coupled chromatography and radiography to demonstrate that 106Ru has been incorporated into the isolated complex (i.e., [Ru(ttpy)2](PF6)2 + [106Ru(ttpy)2](PF6)2). Here, we have used thin-layer chromatography (TLC) (normal phase, preparative scale) to demonstrate the purity of the isolated [Ru(ttpy)2](PF6)2 by elution with an aqueous acetonitrile solution of high ionic strength. The decay progeny, 106Rh [t1/2 = 30 s, β(mean) = 1,410 keV] is a hard beta-particle emitter, making it well-suited to autoradiographic imaging. Prolonged exposure of the phosphor imaging plate (6 wk) yields darkened areas that correlate to the [Ru(ttpy)2](PF6)2 isolated from reactions involving 106Ru-contaminated filter pieces and their aqueous extracts (). As with all radiochemical experiments described herein, every effort to provide a full account of the initially measured 106Ru activity has been made by measuring all subsequently isolated reaction products, by-products, and other relevant materials. In this way, the reactive and unreactive portions of the 106Ru contaminant can be independently determined, where 106Ru localized with [Ru(ttpy)2](PF6)2 constitutes the reactive portion and 106Ru measured in other materials and by-products (AgCl) constitutes the unreactive portion. For all reactions, a reasonably full account has been provided in consideration of the respective measurement uncertainty carried by each experimental component (General) and of the potential loss of material given the considerable number of mass transfer steps involved (Radiochemistry).

Direct Reaction with a Contaminated Filter Piece.

Direct reaction of a portion (0.144 g, 7.99 ± 0.53 Bq) of the 106Ru-contaminated filter piece in the presence of β-RuCl3 resulted in 79.6 ± 4.9% (6.36± 0.39 Bq) of the 106Ru being localized with the isolated [Ru(ttpy)2](PF6)2 after gamma counting. The balance of the activity (unreactive portion) was reasonably well accounted for, with 8.62 ± 0.68% (0.689 ± 0.054 Bq) and 7.77 ± 0.60% (0.460 ± 0.035 Bq) measured in the remaining filter piece and the insoluble AgCl by-product, respectively.

Reactions Subsequent to Aqueous Extraction of a Filter Piece.

To gain a bit more resolution, a similar reaction was performed on the aqueous extract of another filter piece. The 106Ru that partitioned into the water extract (51.3 ± 2.4%, 6.14 ± 0.46 Bq) was found to be almost completely localized in the isolated [Ru(ttpy)2](PF6)2 (95.3 ± 6.0%, 5.85 ± 0.37 Bq) with the balance of activity reasonably accounted for in the AgCl precipitate (6.06 ± 0.41%, 0.372 ± 0.025 Bq).

The filter piece was found to retain 27.8 ± 1.9% (3.33 ± 0.22 Bq) of the 106Ru contaminant after aqueous extraction. The balance of 106Ru activity (∼20.9%) was most likely lost in transfer to the submicron filtration material which could not be accommodated in the well-detector space for measurement (). Direct reaction of the water-washed filter piece resulted in 75.2 ± 4.9% (2.50 ± 0.16 Bq) of the activity being localized in the isolated [Ru(ttpy)2](PF6)2, with the balance reasonably accounted for in the AgCl by-product (16.2 ± 1.1%, 0.540 ± 0.038 Bq) and the remaining filter piece (9.06 ± 0.71%, 0.302 ± 0.024 Bq). Assuming that the activity remaining on the filter piece is representative of the entirety of 106Ru that did not partition into the aqueous phase (i.e., 48.7 ± 2.4%) and summing reactive and unreactive contributions, we find that 85.6 ± 8.1% and 15.2 ± 1.4% of the initial 106Ru contaminant to be reactive and unreactive, respectively. These results are in reasonably good agreement with those obtained by direct reaction on a 106Ru-contaminated filter piece. Clearly, one or more of the water-insoluble components is chemically inert to the reaction conditions employed. According to Table 2, chemical inertness and a high degree of water insolubility are certainly consistent with the occurrence of RuO2. Against the previously mentioned compositional characteristics for both trapping and pyroprocessing scenarios, the delineation of a small Ru(III) polychlorinated component (1.17 ± 0.36%) and a substantial water-insoluble, reactively inert component (∼20%) consistent with RuO2 support an origin from the reductive trapping of RuO4 in HCl. Such activity further explains the high radiopurity of radio-ruthenium observed in environmental samples (1), since volatilization of RuO4 from waste streams stands alone as the means to provide optimal separation efficiency (14).

Concluding Remarks

We have revealed compositional markers for the 106Ru contaminant that are uniquely consistent with nuclear waste reprocessing, namely the reductive trapping of RuO4 in HCl. This finding aligns with other empirical evidence pertaining to radiopurity and age estimation (i.e., time since removal from irradiation) for spent nuclear fuel obtained from measurement of the 103Ru:106Ru ratio (1). An age estimate of ∼2 y is considerably less than the usual time (≥3 y) allotted before reprocessing under typical reactor operating conditions (52). Therefore, a RuO4 trapping process in HCl would undoubtedly be exothermic, exacerbating an already energetic mixture that, in light of the high volatility and potential explosive decomposition of RuO4 (15), gives credence to the occurrence of an abrupt, uncontrolled release (i.e., explosion). Such conditions would certainly aid in the volatilization and dispersion of otherwise nonvolatile ruthenium species. For a facility undertaking the purification of fission products from spent nuclear fuel for commercial gain, it is hard to ignore the potential profit afforded by obtaining 106Ru in high-specific activity, considering its long-standing medical use in the development of brachytherapy plaques for the treatment of eye cancer (53, 54). For this application, it is commonly electrodeposited from solution onto silver, along with the addition of carrier RuCl3 (54, 55). Since this requires a highly soluble form of 106Ru, it makes sense to invoke a reductive technique that generates polychlorinated Ru(III/IV) species with a high degree of radiopurity and aqueous solubility, rather than the vast majority of passivation techniques that ultimately produce the highly insoluble RuO2.

In closing, the detection of 106Ru in aerosol filters across European surveillance networks in 2017 represents a limiting case for forensic investigation, considering the high radiopurity of the contaminant and the absence of detectable signatures or anomalies from colocated stable elements. However, as this work has demonstrated, new opportunities arise when the radioisotope pertains to an element capable of covalent bond formation. By subjecting the radioactive contaminant to iterative, well-characterized chemical transformations and determining its subsequent fate and distribution, we create an inferential process from which we can gain both generalized and highly specific information about the chemical form(s) of the radioactive contaminant. This work constitutes direct evidence for specific nuclear fuel reprocessing activity and, coupled with other measurements and atmospheric dispersion modeling, provides irrefutable proof as to the origin of the 2017 environmental release of 106Ru. Moreover, this work serves, by example, as a potentially valuable addition to the established suite of nuclear forensic capabilities.

Materials and Methods

General.

Particulate filters containing 106Ru from the radiological monitoring networks of Germany (German Meterological Services Deutscher Wetterdienst) and Sweden (Swedish Defense Research Agency Totalförsvarets forskninginstitut) were graciously donated. Chemical experimentation was conducted on portions of a particulate filter sample obtained from Vienna, Austria, and was composed of polypropylene (air collection from 2017-09-28 to 2017-10-04 at a rate of 675 m3/h).

Chromatographic supports consisting of alumina (neutral, type WN-6, super grade, flash chromatography), silica (high-purity grade, 220 to 440 mesh, flash chromatography), and preparative TLC plates (glass-backed, 2.0-mm SiO2 layer) were purchased from Sigma-Aldrich, Ltd. Columns (C18, XBridge: 3.5 µm, 250 × 4.6 mm and 5 µm, 250 × 10 mm) were purchased from Waters, Ltd. for analytical and preparative-scale HPLC separations, respectively. All aqueous and ethanolic extracts of contaminated filter pieces were filtered through conditioned syringe filters (0.1-µm pore size; Fisher Scientific) prior to gamma counting and subsequent reaction.

Solvents and reagents were used as received. These include N,N′-dimethylformamide (Fisher Chemical, ACS grade), acetonitrile (Fisher Chemical, ACS grade), methanol (Fisher Chemical, Optima grade), dichloromethane (Fisher Chemical, ACS grade), diethyl ether (Fisher Chemical, ACS grade), potassium nitrate (Sigma-Aldrich, ≥ 99%), silver nitrate (Sigma-Aldrich, ≥ 99%), ammonium hexafluorophosphate (Acros Organics, 99%), 2-acetyl pyridine (Sigma-Aldrich, 99%), p-tolualdehyde (Sigma-Aldrich, 97%), β-ruthenium chloride hydrate (Sigma-Aldrich, reagent plus), ammonium hexachlororuthenate (Strem Chemicals, 99%), ruthenium (III) nitrosyl nitrate solution (Sigma-Aldrich, 1.5% Ru in 6.8 wt % nitric acid), ruthenium (IV) oxide (Sigma-Aldrich, 99.9%), potassium perruthenate (Sigma-Aldrich), α-ruthenium chloride (Merck), and tetrabutylammonium chloride (Sigma-Aldrich, 98%).

Experimental uncertainty was derived by combining contributions from counting statistics, counting efficiency, and mass transfer (if applicable) in quadrature.

Synthesis and Characterization.

Synthesis of 4′-p-tolyl-2,2′; 6′,2″-terpyridine (ttpy).

Synthesis of this ligand resembled previously reported procedures (35, 36); however, these procedures were found to yield unsatisfactory purity considering the application. In our hands, 2-acetylpyridine (7.83 g, 0.065 mol and KOH aqueous solution (5 mL, 15 wt %) were stirred briefly in methanol (60 mL) at room temperature (∼5 min). p-tolualdehyde (3.56 g, 0.03 mol) and concentrated ammonium hydroxide (25 mL) were then added and the mixture heated to reflux with vigorous stirring for 48 h. After cooling, the reaction mixture was decanted into a large separatory funnel. To this was added 600 mL of water and 600 mL of dichloromethane. After agitation, the dichloromethane layer was removed, washed once more with 600 mL water, then separated and dried over sodium sulfate. The dichloromethane was removed by distillation and the remaining residue was recrystallized from 95:5 ethanol/water to yield 2.45 g of slightly impure material. High purity was achieved by flash chromatography using 688 g of alumina, previously deactivated by thorough mixing with water (5% by mass) and using toluene as the mobile phase. Yield = 1.91 g (20%); 1H NMR (600 MHz, d6-DMSO) δ ppm: 2.40 (s, 3H, Htolyl), 7.40 (d, J = 8.0 Hz, 2H, Htolyl 3,5), 7.53 (dd, J = 7.5 Hz, J = 5.0 Hz, 2H, H5,5′′), 7.83 (d, J =8.0 Hz, 2H, Htolyl 2,6), 8.04 (t, J = 7.5 Hz, 2H, H4,4″), 8.67 (d, J = 8.0 Hz, 2H, H3,3′′), 8.70 (s, 2H, H3′,5′), 8.76 (d, J = 5.0 Hz, 2H, H6,6″). ESI(+)MS found (calcd) for C22H17N3H: 324.1511 (324.1501).

Synthesis of ttpyRuCl3.

This compound was prepared according to literature procedure. Note that reactant concentration should be maintained ≥0.01 M to attain high yield. Note also that the ethanol extract from a filter piece contaminated with 106Ru, after filtration through a submicron filter, would be used in place of ethanol alone, according to the following typical preparation. The ligand ttpy (0.0825 g, 2.55 × 10−4 mol) and RuCl3 hydrate (0.0607 g, 2.32 × 10−4 mol) were combined in 20 mL of ethanol (95%) and heated to reflux with agitation for 2 h. After cooling, the insoluble solid was isolated by filtration over a glass filter frit then agitated 5 to 10 min in ethanol (40 mL) and filtered. The residue was then rinsed with an additional portion of ethanol (40 mL) and finally with diethyl ether (40 mL). The solid was dried under vacuum. Yield = 0.123 g (88%). ESI(+)MS found (cald) from DMF solution for C44H34N6RuCl3Na: 554.9445 (554.9416).

Synthesis of [Ru(ttpy)2](PF6)2 from β-RuCl3.

Procedure (A).

This compound was prepared similarly to literature procedures. Typically, the ligand ttpy (0.111 g, 3.42 × 10−4 mol, 2.1 equivalents), RuCl3 hydrate (0.0426 g, 1.63 × 10−4 mol), and AgNO3 (4.89 × 10−4 mol) were combined in N,N′-DMF (reagent grade, 50 mL) and heated to reflux for 16 h. The reaction mixture was then cooled to room temperature, and the AgCl by-product was removed by vacuum filtration over a glass filter frit. The filtrate was then distilled to dryness and the residue chromatographed on silica using an acetonitrile/saturated, aqueous potassium nitrate (7:1) mixture as mobile phase. The isolated compound was transferred to a separatory funnel, followed by addition of water, NH4PF6, and enough dichloromethane to render a phase separation. After washing, the aqueous layer was removed and discarded. This process was repeated twice more, after which the organic phase was collected and distilled to dryness. The residue was then redissolved in acetonitrile and precipitated from water. This precipitate was collected by vacuum filtration, redissolved in acetonitrile, and precipitated from diethyl ether. This final precipitate was collected by filtration and dried under vacuum. Yield = 0.160 g (94%); 1H NMR (600 MHz, CD3CN) δ ppm: 2.54 (s, 3H, Htolyl), 7.18 (dd, J = 8.0 Hz, J = 6.0 Hz, 2H, H5,5′′), 7.42 (d, J = 6.0 Hz, 2H, H6,6″), 7.58 (d, J = 8.0 Hz, 2H, Htolyl 3,5), 7.94 (t, J = 8.0 Hz, 2H, H4,4″), 8.11 (d, J = 8.0 Hz, 2H, Htolyl 2,6), 8.64 (d, J = 8.0 Hz, 2H, H3,3′′), 8.99 (s, 2H, H3′,5′). ESI(+)MS found (calcd) for C44H34N6Ru (M2+): 373.9200 (374.0962).

Procedure (B).

Alternatively, the ligand ttpy (0.111 g, 3.42 × 10−4 mol, 2.1 equivalents) and RuCl3 hydrate (0.0426 g, 1.63 × 10−4 mol) were combined in ethanol (95%, 50 mL) and heated to reflux for 16 h. Distillation of the ethanol, followed by chromatographic isolation and work-up as outlined above in Procedure (A) gave the target complex in 89% yield (0.151 g). Notably, the same molar quantities and conditions applied to (NH4)2RuCl6 and Ru(NO)(NO3)3 gave comparable yields (94 and 90%, respectively) for the formation of [Ru(ttpy)2](PF6)2.

Synthesis of [Ru(ttpy)2](PF6)2 from α-RuCl3, (NH4)2RuCl6, Ru(NO3)3(NO), KRuO4, and RuO2 in N,N′-DMF.

These compounds were reacted, and subsequently isolated, according to the scale and conditions outlined in Procedure (A) for β-RuCl3. Yield (α-RuCl3) = 0.162 g (95%); yield ((NH4)2RuCl6) = 0.163 g (96%); yield (Ru(NO3)3(NO)) = 0.156 g (92%); yield (KRuO4) = 0.032 g (19%); yield (RuO2) = no reaction.

Gamma Spectrometry.

The gamma detection system used was a small anode germanium (SAGe) well detector (GSW275L; Mirion Technologies) outfitted with a cosmic veto (plastic scintillator) detector to reduce the background (CV System-LM; Mirion Technologies). The SAGe well detector has a diameter of 28.00 mm and a depth of 40.00 mm with an active volume for the germanium crystal of 65.50 mm (thickness) by 86.6 mm (diameter). The resolution of the detector (full width at half maximum) is 1.835 keV at 1,332.5 keV. The gamma acquisition software used was Genie 2000 v3.4.1 (Mirion Technologies), with the counting efficiencies of the samples in the well of the SAGe simulated using LabSOCS v4.4.1 (Mirion Technologies). The analysis software used in quantifying the activity of 106Ru was a peak-fitting and peak identification program called UniSAMPO (v 2.67)-Shaman (v 1.2), developed by Baryon Oy of Finland. The HPGe well detector was calibrated for energy and shape resolution using a National Institute of Standards and Technology–traceable standard (SRS 112559; Eckert & Ziegler) prior to measurement. Data were collected from 0 to 2,800 keV for 16,384 channels. Samples were counted in 20-mL glass scintillation vials (61 × 28 mm, outer dimensions). Detector count times varied greatly depending on the amount of 106Ru in the sample and ranged anywhere from several thousand to several million seconds, ideally until an acceptable counting uncertainty was attained (<10%). Analysis was based upon the 622-keV gamma emission (9.93% abundance) associated with the 106Ru progeny, 106Rh, while decay corrections were performed using t1/2 = 371.8 d for 106Ru. For the experiment incorporating 106Ru into ttpyRuCl3, the purified reaction components were gamma counted until the critical limit (Lc) was exceeded and the peak was automatically identified by the peak search algorithm of the analysis software (UniSAMPO-Shaman, Baryon Oy, Finland). For instance, detection count times ranging from 1 to 2 Ms corresponded to detection capabilities, as defined by the Lc and detection limit (Ld) at 95% confidence, of Lc = 3.7 mBq, Ld = 7.5 mBq and Lc = 2.9 mBq, Ld = 5.9 mBq, respectively.

NMR and Mass Spectrometry.

The 1H NMR spectra were acquired on a Bruker AVANCE III 600 MHz spectrometer by the University of Ottawa NMR Facility. Chemical shifts are reported relative to Me4Si as an internal reference. Mass spectra were obtained using a Micromass Q-TOF II Electrospray Ionization Mass Spectrometer by the John L. Holmes mass spectrometry facility at the University of Ottawa.

HPLC.

HPLC was performed using a Dionex ICS-6000 instrument equipped with photodiode array detector (PDA-1; Thermo Scientific), autosampler (AS-AP; Thermo Scientific), and fraction collector (ASX-280-FC; Thermo Scientific) and employing a C18 column (3.5 µm, 250 × 4.6 mm and 5 µm, 250 × 10 mm) from Waters, Ltd. Chromeleon 7 (Thermo Scientific) was the software package used for instrument control and analysis. All injections and subsequent runs were monitored at both 254- and 450-nm wavelength.

Radiochemistry.

Transformation of β-RuCl3 to form either ttpyRuCl3 or [Ru(ttpy)2](PF6)2 in the presence of 106Ru adhered to the respective protocols outlined above in . Typically, a filter piece was shaved with a razor into many thin pieces, and these were placed inside a scintillation vial for gamma counting prior to reaction. Once transferred to the reaction vessel, the empty scintillation vial was gamma-counted once more to ensure the efficacy of transfer.

Reactions with β-RuCl3.

For reaction to form ttpyRuCl3 (), the shaved filter pieces (1.03 ± 0.08 Bq 106Ru) were immersed in 5 mL of ethanol and agitated in an ultrasound bath for 10 to 20 min, after which the ethanol was removed via syringe and filtered. The process was repeated twice more, and the ethanol fractions were combined and gamma-counted (0.074 ± 0.005 Bq 106Ru). The ethanol washings were then transferred to a reaction vessel, along with an additional 5 mL of ethanol (rinse). The reaction protocol described herein for ttpyRuCl3 () was carried out, giving the target complex in 88% yield. This material was then gamma-counted, as were the subsequent components from its HPLC purification.

For direct reaction of a filter piece to form [Ru(ttpy)2](PF6)2 (), the gamma-counted filter shavings (7.99 ± 0.53 Bq 106Ru) were transferred to a reaction vessel followed by the addition of 10 mL of water. After brief agitation in an ultrasound bath, the reaction protocol outlined herein for [Ru(ttpy)2](PF6)2 was carried out (), giving a comparable yield for the final product (93%). All components (i.e., [Ru(ttpy)2](PF6)2, AgCl by-product, and the remaining filter pieces) were subsequently gamma-counted.

For reaction to form [Ru(ttpy)2](PF6)2 from the aqueous extract (), the shaved filter pieces (11.97 ± 0.89 Bq 106Ru) were immersed in 5 mL of water and agitated for 10 to 20 min in an ultrasound bath, after which the water was separated via syringe and filtered. The process was repeated twice more, and the aqueous fractions were combined and gamma-counted. The aqueous extract was then transferred to a reaction vessel, followed by an additional 5 mL of water (rinse). Complexation was carried out according to the procedure described herein for [Ru(ttpy)2](PF6)2 (), with the exception that a higher-proportion of DMF was used (60 mL). The target complex, [Ru(ttpy)2](PF6)2, was isolated in 95% yield. All components (i.e., [Ru(ttpy)2](PF6)2, AgCl by-product, and the remaining filter pieces) were subsequently gamma-counted. The remaining, washed filter pieces were reacted separately in DMF to give the isolated [Ru(ttpy)2](PF6)2 in 94% yield, which was then gamma-counted along with the other reaction components.

Carrier-free reaction.

Shaved filter pieces (0.48 ± 0.03 Bq 106Ru) were immersed in 5 mL of ethanol and agitated in an ultrasound bath for 10 to 20 min, after which the ethanol was removed via syringe and filtered. The process was repeated twice more, and the ethanol fractions were combined and gamma-counted. The ethanol washings were concentrated by vacuum distillation then transferred with rinse solutions to a 3-mL conical reaction vessel. The ligand (0.20 mg, 6.18 × 10−7 mol) was added from a stock solution in ethanol to give a final reaction volume of 0.5 mL. A stir bar was added and a Teflon screw cap was secured. The reaction solution was agitated while immersed in an oil bath set to 90 °C for 16 h (). After cooling to room temperature, the stir bar was removed and the reaction solution transferred to a distillation flask, followed by 20 mg of [Ru(ttpy)2](PF6)2 (stable Ru). The solvent was removed by distillation, and the mixture was chromatographed on silica gel and worked up as previously described above in . The isolated complex was quantitatively recovered and gamma-counted.

Autoradiography.

Autoradiography of chromatographed reaction products (), isolated in the presence of 106Ru, was performed on a BAS-5000 Image Analysis System (Fujifilm). Imaging plates (BAS-SR2025; GE Healthcare) were developed in closed imaging cassettes (EXPSR CASS. 20 × 25 cm; GE Healthcare) and erased (zeroed) using an IP Eraser 3 (Fujufilm). Images were acquired using BASReader software (Fujifilm) and processed using Multi Gauge v.3.1 (Fujifilm).

Data Availability.

All relevant data and protocols (synthetic and radiochemical) are provided in the paper and .