An unknown source of reactor radionuclides in the Baltic Sea revealed by multi-isotope fingerprints

The study investigates whether previously unrecognized sources of anthropogenic uranium isotopes exist in the Baltic Sea beyond known inputs from European nuclear fuel reprocessing (La Hague and Sellafield) and global fallout from atmospheric nuclear weapons testing. Uranium-236 (236U) is a sensitive tracer for nuclear fuel cycle releases, but its source apportionment is complicated by ubiquitous fallout and the dominance of natural 238U. The key hypothesis is that combining multiple isotopic fingerprints—particularly the diagnostic 233U/236U ratio, which is high in weapons fallout and very low in reactor-related releases—together with 236U/238U, 236U/129I, and 129I/127I, can reveal and localize unknown reactor-derived contributions to the Baltic Sea. The work is motivated by environmental and public health concerns given the Baltic’s limited water exchange and documented radionuclide pollution, and aims to improve nuclear safeguards, emergency preparedness, and tracer-based oceanographic studies.

Prior studies have traced reactor-derived 236U and 129I from reprocessing plants at La Hague (France) and Sellafield (UK) throughout the North Atlantic–Arctic system and into the North Sea. Global fallout contributed an estimated >1000 kg of 236U to Earth’s surface, complicating identification of reactor sources when using only 236U/238U. The 233U/236U atomic ratio provides a powerful discriminator: global fallout is characterized by ~1.4 × 10−2, whereas reactor releases are orders of magnitude lower (~10−7–10−6 for La Hague discharges), consistent with reactor physics. In the Irish Sea, dominated by Sellafield discharges, 233U/236U averages ~0.12 × 10−2, reflecting a reactor signature. Previous work documented 129I distributions in the North Sea–Skagerrak–Kattegat and minor Baltic riverine 129I contributions, enabling dual-tracer approaches (129I with 236U) to track advection pathways from reprocessing sources. However, comprehensive multi-isotope (233U–236U–129I) surveys of the Baltic Sea to resolve local sources had been lacking.

- Study area and sampling: Surface (0–5 m, some deeper) seawater and surface sediments were collected across the Baltic Sea and adjacent waters (western Danish coast, Kattegat–Skagerrak) during 2011–2016, plus one lake water sample (Lake Mälaren). Sampling locations were grouped into five regions: KGR (Kattegat–Skagerrak and Danish west coast), DS (Danish Straits), SBR (South Baltic), MBR (Middle Baltic), NBR (North Baltic including Bothnian Sea and Bay). Five Baltic surface sediments (2016) and one Studsvik-area sediment (2014) were analyzed. - Chemical preparation: Seawater was filtered, acidified, and subjected to Fe(OH)3 co-precipitation; uranium was purified using UTEVA resin. Sediments were ashed and leached with aqua regia, then processed analogously. - Measurements: 238U and 127I were quantified by ICP-MS. Anthropogenic isotopes 236U/238U and 233U/236U were measured by AMS at VERA (University of Vienna). 129I/127I was measured by AMS at Uppsala University after AgI preparation. Quality control employed in-house standards (Vienna-KkU, Vienna-US8), instrument blanks, and monitoring of potential hydride interferences; reported detection limits for 236U/238U were <1 × 10−14. - Data analysis: Spatial distributions of concentrations and isotope ratios were mapped. Salinity-mixing relationships were assessed: 238U and 129I vs salinity showed strong linear trends indicating two-endmember mixing between saline North Sea water and riverine freshwater. Binary mixing models were developed: • Model L: Best-fit binary mixing between North Sea water and a freshwater endmember carrying only global fallout-derived 236U (no reactor 236U), constrained using observed 233U/236U. • Model L1: Mixing with freshwater containing no 236U (for comparison). • Model L2: Mixing between North Sea water and freshwater characterized by fallout 236U apportioned using 233U/236U to quantify reactor vs fallout contributions. Deviations from mixing lines and their spatial patterns were used to localize additional sources. The excess 236U inventory was estimated from the difference between L and L2 at the mean Baltic salinity, integrating over the Baltic water volume, with uncertainties following robust evaluation procedures.

- Elevated anthropogenic uranium in the Baltic: 236U/238U atomic ratios span (5–52) × 10−9, with higher values in central and northern Baltic regions and lower in the western Baltic (Kattegat–Skagerrak–Danish Straits). The maximum is ~6× the North Sea average in 2010 ((7.6 ± 3.7) × 10−9). - High 236U concentrations: Surface waters in the Bothnian Sea and Bothnian Bay show (6–9) × 10^7 atoms/L, comparable to the central North Sea ((3–10) × 10^7 atoms/L), and average 236U/238U increases threefold from KGR ((10 ± 3) × 10−9) to MBR/NBR ((32 ± 7) × 10−9), indicating additional local 236U inputs. - 233U/236U ratios: Range (0.14–0.87) × 10−2. A Kattegat–Skagerrak subgroup shows ~0.20 × 10−2, consistent with mixing of North Sea water influenced by reprocessing. A large Baltic cluster (median salinity ~6.9‰) shows (0.53 ± 0.03) × 10−2, significantly below a fallout-only mixing prediction, implicating an additional low-233U reactor source. - Iodine distributions: 129I concentrations are (3–232) × 10^9 atoms/L; 129I/127I ratios are (101–1286) × 10−9, highest in North Sea–Skagerrak–Kattegat and decreasing into the Baltic Proper, consistent with reprocessing sources and minor riverine contributions. - Dual-tracer ratios: 236U/129I spans (5–133) × 10−4 and increases more than tenfold from KGR (average (8 ± 2) × 10−4) to the central Baltic (~1 × 10−2). If the local 236U source contains little/no 129I, this implies roughly 90% of 236U in central Baltic surface waters is locally derived. - Mixing and source apportionment: Using 233U/236U to separate endmembers, the reactor-derived 236U contribution is estimated at 2.1 ± 0.2 times the fallout-derived 236U in much of the Baltic surface waters. A freshwater endmember consistent with fallout yields 236U = (3.56 ± 0.39) × 10^7 atoms/L and 236U/238U = (3.52 ± 0.39) × 10−8 in the L2 model. - Inventory estimate: The excess reactor-derived 236U inventory in the Baltic Sea is 200 ± 47 g (snapshot based on surface data and mean salinity). The remaining 95 ± 22 g is attributed to global fallout remaining in Baltic waters under residence-time considerations. - Potential sources and localization: Spatial deviations from mixing indicate sources in the middle and northern basins rather than solely riverine entry points. Sediment near Studsvik (Bergasundet) has very high 236U content ((2.02 ± 0.12) × 10^13 atoms/kg) and a reactor-like 233U/236U ((0.36 ± 0.05) × 10−2), three orders of magnitude higher 236U than other Baltic sediments (which show 233U/236U of 0.59–0.83 × 10−2). Lake Mälaren water shows 236U/238U ~2 × 10−8 and 233U/236U (0.18 ± 0.05) × 10−2, consistent with reactor signatures, but its flux implies negligible 236U delivery (~0.1 g/yr). Reported Westinghouse 236U releases (0.44 g over 1998–2017) are far too small to account for the observed excess. - Alternative source scenario: Leakage/dissolution from dumped/damaged spent nuclear fuel on the seafloor could also explain reactor-like 236U with minimal 233U; typical fuel can have 236U/238U up to ~1 × 10−2, so relatively small amounts of fuel (tens of kg) could account for the observed inventory.

The multi-isotope approach definitively shows that known sources (North Sea inflow from reprocessing plants and global fallout) cannot alone explain Baltic Sea 236U patterns. The elevated 236U/238U and especially the intermediate 233U/236U ratios in central and northern basins indicate a substantial addition of reactor-derived 236U with negligible 233U. The 236U/129I increase inland further supports a local 236U source with little 129I co-release. Binary mixing constrained by 233U/236U quantifies reactor-derived 236U at roughly double the fallout contribution and localizes the source to interior basins, inconsistent with simple riverine dilution. The geospatial fingerprint, together with extraordinarily high 236U in sediments near Studsvik and reactor-like 233U/236U there, points to historical or ongoing discharges from Swedish nuclear research activities as a plausible source. Alternatively, leakage from dumped or lost spent nuclear fuel could contribute; such a source would raise concerns about broader radionuclide releases (e.g., 137Cs). These findings demonstrate the value of 233U/236U as a sensitive diagnostic for environmental nuclear forensics and underscore the need for improved records and monitoring of regional nuclear facilities and potential seabed disposal sites.

This study applies a combined 233U–236U–129I fingerprinting strategy to reveal an unknown reactor-derived source of 236U in the Baltic Sea. Surface waters show elevated 236U/238U, 236U/129I, and characteristic 233U/236U ratios indicating that about two-thirds of anthropogenic 236U originates from a local reactor source rather than from global fallout or North Sea inflow. A first-order inventory suggests ~200 ± 47 g of reactor-derived 236U currently resides in Baltic waters. Spatial patterns and sediment evidence implicate discharges near Studsvik as a plausible contributor, though leakage from seabed nuclear materials cannot be excluded. The results highlight the diagnostic power of the 233U/236U signature for nuclear safeguards and environmental tracing. Future work should include depth profiles, time-series monitoring, comprehensive sediment–water coupling assessments, improved source term documentation from regional nuclear installations, and targeted surveys of potential seabed disposal or dumping locations to better constrain sources, transport, and risks.

- Inventory estimates are first-order and based mainly on surface waters from multi-year sampling; vertical distributions, deep-water inventories, and interannual variability are not fully constrained. - Uranium scavenging in anoxic regions and sediment–water interactions could alter concentrations though not isotope ratios; these processes introduce uncertainty in mass balance estimates. - Salinity patterns and North Sea inflow events vary and may affect mixing analyses. - The contribution of Chernobyl-derived 236U is difficult to quantify and was not resolved explicitly. - Limited or missing release records for regional nuclear facilities hinder precise source attribution. - 129I analyses were only available for samples collected in 2015, reducing spatiotemporal coverage for iodine-based mixing constraints.