Non-natural ruthenium isotope ratios of the undeclared 2017 atmospheric release consistent with civilian nuclear activities

Understanding the circumstances of the undeclared 2017 nuclear release of ruthenium that led to widespread detections of the radioisotope 106Ru in the Eurasian region, and whether it derives from a civilian or military source, is of major importance for society and future improvements in nuclear safety. Until now, the released nuclear material has merely been studied by analyzing short-lived radioisotopes. Here, we report precise measurements of the stable isotopic composition of ruthenium captured in air filters before, during, and after the nuclear release, and find that the ruthenium collected during the period of the 2017 nuclear release has a non-natural isotopic composition. By comparing our results with ruthenium isotopic compositions of spent nuclear fuels, we show that the release is consistent with the isotopic fingerprints of a civilian Russian water-water energetic reactor (VVER) fuel at the end of its lifetime, and is not related to the production of plutonium for nuclear weapons.


Supplementary Note 1. Analyses
To estimate the true external reproducibility of used air filter samples, standard deviation (2 s.d.) of the Ru isotope ratios determined for the reference air filters can be calculated (Supplementary Table 3). This more conservative uncertainty estimate using the standard deviation (2 s.d.) on the Ru isotope ratios of all four reference air filters may include some natural variations of the Ru isotopic composition in the atmospheric Ru over time (i.e., atmospheric Ru may partly originate from vehicle catalytic converters that may have led to isotope fractionation). However, the overall good agreement of all four filter samples collected from 2015 to 2018 can be used to describe and characterize the general isotopic composition of "background Ru" in the atmosphere over Vienna, Austria (Supplementary Table 3 The air filter containing the non-natural Ru from the nuclear release in 2017 could also be analyzed twice, using 10 ppb and 1 ppb solutions, respectively. The two measurements agree very well with each other and reveal highly non-natural Ru isotopic composition of the Ru collected by this air filter for all Ru isotope ratios (Supplementary Table 3).

Supplementary Note 2. Isotopic composition of fission-generated Ru
Supplementary Table 4 summarizes the Ru isotopic composition of fission-generated Ru from different reactor types and the thermal neutron fission yields expected for low-burnup 239 Pu production. These values were derived from actual measurements of nuclear fuels, nuclear waste, contaminated ground water, or calculated based on fission product yields [1][2][3][4] . Fission-generated Ru does not have significant abundances of 96 Ru, 98 Ru, and 99 Ru, because long-lived   or stable molybdenum and technetium isotopes inhibit the beta-decay along the corresponding isobars, respectively. Artificial generation of these lighter Ru isotopes are possible at trace levels only due to (rather improbable) independent fission events yielding these Ru isotopes directly (or by the extremely slow decay of 99 Tc, respectively). In contrast, the heavier Ru isotopes ( 101 Ru, 102 Ru, and 104 Ru) are produced in higher abundances than the natural abundances because their production path along the isobar is not blocked by any long-lived or stable isotope. The isotopic deviation of fission-generated Ru vs. naturally abundant Ru is greatest for the isotope 101 Ru (Supplementary Table 4). Hence, the i Ru/ 101 Ru ratios can be used to distinguish between various origins of fission-generated Ru and thus were favored for the illustration of the nuclear impact on the isotopic Ru composition in Fig. 1.
Available data of 102 Ru/ 101 Ru in low-burnup nuclear fuel 2 and the Hanford Site groundwater 1 allow deriving an estimate for the expected Ru isotopic composition generated during the production of weapons-grade Pu (Supplementary Table 4). This estimate is in good agreement with the calculated 102 Ru/ 101 Ru of 235 U thermal neutron fission yields, with minor contributions of 239 Pu thermal neutron fission (Supplementary Table 4). The estimated 102 Ru/ 101 Ru ratio for Ru produced during 239 Pu production is different from any of the 102 Ru/ 101 Ru signatures of civilian reactors (Supplementary Table 4). Moreover, during low-burnup and thermal neutron fission of 235 U or 239 Pu, no significant amount of 100 Ru is produced, hence, the 100 Ru/ 101 Ru ratio of Ru from civilian reactors is also a distinctive signature for tracing the origin of fissiongenerated Ru (Supplementary Table 4).

Supplementary Note 3. Ru isotopic fingerprints of various reactor types
In preparation of the German nuclear waste repository, the 'Gesellschaft für Anlagen-und Reaktorsicherheit' (GRS) 4 modelled and published the isotopic inventory of all spent nuclear fuel from all reactor types that have ever been in operation in Germany. The link https://www.grs.de/en/node/1749 will bring you to the root menu, from where the tabulated data can be found under "Anhang GRS-278." For this study, we used "Waste from the utilization of power reactors" (1 Abfälle aus der Nutzung von Leistungsreaktoren), "Irradiated nuclear fuel elements" (Bestrahlte Brennelemente), "Nuclear fuel" (Kernbrennstoff). The reactor types used for this study are "Pressurized water reactors" with both MOX and UO2 as fuel (DWR-MOX and DWR-UO2, respectively), Boiling water reactors" (SWR-MOX and SWR-UO2, respectively), VVER-210 (KKR), and VVER-440 (KGR).
The enrichment and burnup parameters used in the GRS modelling are tabulated in Supplementary Table 5. Chemical impurities are listed in Supplementary Table 6. Fuel cladding was Zircaloy 4. Nuclear cross sections were taken from the library ENDF/B VI.

Supplementary Note 4. Estimate of non-natural Ru fraction
The non-natural Ru isotopic composition in the air filter can be explained by mixing of fissiongenerated Ru from the undeclared atmospheric release in 2017 with natural Ru. The determined isotopic composition falls closest to mixing lines of the Ru expected to be produced by a VVER reactor and natural Ru. The fraction of fission-generated Ru required can be estimated by massbalance calculations using the calculated Ru isotope abundances (Supplementary Table 7) and is, depending on the Ru isotope used in the calculation, ~66-90 % (excluding 104 Ru due to possible larger interference effects from 104 Pd). This means that 10 to 34 % of the Ru analyzed in the sample solution was natural atmospheric Ru collected from the air or added as analytical blank during digestion and chemical separation (despite using ultra-high-purity chemicals). In any case, the majority of the Ru analyzed in this filter sample was generated in a nuclear reactor.
To better estimate the absolute amount of Ru that was collected by the air filter station in Vienna, Austria, during the week of the 2017/09/28 to 2017/10/04, we analyzed a small aliquot (2 %) of the sample solution for its Ru concentration using a ThermoScientific X-Series 2 quadrupole ICPMS at the Institut für Planetologie in Münster. The measured intensity of the sample solution was compared to a standard solution to estimate the total amount of Ru in the sample aliquot. To calculate a true estimate of the Ru content, the non-natural isotopic composition of the Ru in the air filter has to be taken into account (Supplementary Table 7).
After correction for the non-natural isotopic composition, the amount of Ru in the sample solution is estimated to be ~5 ng (±20 %). Hence, during chemical separation and removal of Mo and Pd interferences, up to 50 % of the Ru was lost. If we assume that ~66 to 90 % of the total Ru trapped on the filter is fission-generated, it translates to 3-4.5 ng in the ~2.5 g filter material digested (Supplementary Table 1). Since the fission-generated Ru was homogeneously distributed (as shown in 5 ), we can estimate the total amount of fission-generated Ru collected in the complete filter (total mass 9.7 g) to be 11.6-17.5 ng. The total amount of Ru released into the atmosphere during the undeclared release is not well known, but can be estimated to ~110 g, based on the Ru isotopic composition of spent fuel and the 106 Ru source term of 250 TBq. In combination with these estimations, a fraction of ~1x10 -10 of the total released stable Ru and

Supplementary Tables
Supplementary Table 1