Chemical characterisation of degraded nuclear fuel analogues simulating the Fukushima Daiichi nuclear accident

Following the March 11, 2011 Fukushima Daiichi loss-of-coolant accident, core melts interacted with structural materials and concrete to form highly heterogeneous molten core–concrete interaction (MCCI) products. Actual MCCI debris from Units 1–3 has not yet been retrieved due to extreme radioactivity, so laboratory simulants are required to understand phase assemblages, elemental distribution, and actinide redox/speciation relevant to retrieval, storage, and disposal. This study synthesises low-radioactivity MCCI analogues using depleted U and Ce as a Pu surrogate (Nd representing trivalent rare earth fission products) to investigate: (i) what crystalline and glassy phases form, (ii) how U and Ce distribute among phases, and (iii) their local oxidation states under reducing conditions mimicking the accident. Multi-modal micro-focus X-ray methods (μ-XRF/μ-XAS/μ-XRD) are employed to resolve chemistry and mineralogy at micrometre scales.

Large-scale MCCI simulant programs (e.g., VULCANO, VESTA) and thermodynamic modelling have predicted key phases (e.g., (U,Zr)O₂, zircon, anorthite, wollastonite) and microstructures under Fukushima-relevant conditions, but are costly, hazardous, and have not explored Pu or its surrogates extensively. Prior degraded fuel simulants typically added Zr as Zr or ZrO₂, whereas during an accident U–Pu–Zr–O exist as solid solutions. Ce is widely used as a non-radioactive Pu surrogate due to comparable oxidation states and ionic radii, though Ce⁴⁺ reduces to Ce³⁺ more readily than Pu⁴⁺. Thermodynamic analyses for silica-rich concretes suggest zircon formation at longer timescales, while rapid quenching favors anorthite and wollastonite. Earlier micro-focused studies of Chernobyl LFCM simulants showed U predominately U⁴⁺ with local variations in glasses versus crystals, supporting the use of μ-X techniques for highly radioactive, small samples.

Simulant compositions: Three MCCI batches (MCCI-1, -2, -3) maintained constant concrete (SiO₂, CaO, Al₂O₃) and stainless steel components (Fe₂O₃, 316 SS filings), varying the addition route and proportions of U, Zr, Ce, Nd via (U₁−x−yCexZry)O₂−δ solid solutions and/or added ZrO₂. Batches excluded highly radioactive fission products and B₄C. Batched mol% examples (Table 1 in paper) indicate total U 28.04–31.12, total Ce 1.22–1.73, total Zr 6.71–11.35. Synthesis of (U,Zr,Ce,Nd)O₂ solid solutions: Ammonium hydroxide co-precipitation using ZrOCl₂·8H₂O and UO₂(NO₃)₂·6H₂O dissolved in 1 M HCl; precipitation with NH₄OH to pH 9; filtering, washing (water/isopropanol), drying at 60 °C; calcination 750 °C 4 h in N₂/5% H₂; milling, pressing 6 mm pellets; sintering at 1700 °C for 8 h under N₂/5% H₂. MCCI simulant synthesis: Mixed reagents milled (30 Hz, 10 min), sintered in alumina crucibles under reducing atmosphere (5% H₂/95% N₂) at 1500 °C for 4 h; secondary dwell at 720 °C for 72 h to promote crystal growth; heating/cooling rate 3 °C min⁻¹. Bulk characterisation: Powder XRD (Bruker D2 Phaser, Cu Kα, 30 kV/10 mA), 20° ≤ 2θ ≤ 100°, step 0.02°, 0.3 s/step. SEM (Hitachi TM3030, 15 kV, BSE) with Bruker Quantax 70 EDS; samples mounted in epoxy and polished. Micro-focus X-ray analysis: Swiss Light Source, microXAS-X05LA beamline. Si(111) double-crystal monochromator; beam spot ~1 μm × 1 μm at 18,100 eV via KB mirrors. Samples mounted on 250 μm fused quartz slides, thinned to ~50 μm. μ-XRD with DECTRIS Eiger 4M detector over 3.5–42.6° 2θ, Δ2θ ≈ 0.02°, λ = 0.6854 Å. Simultaneous μ-XRF (Si drift detector). 2D μ-diffraction patterns azimuthally integrated and summed (XRDUA) to 1D; automated indexing and phase maps via in-house MATLAB. XANES/EXAFS: U L₃-edge μ-XANES 17,060–17,325 eV (0.2 eV resolution), Y foil (17,038 eV) for calibration; four repeats averaged. U oxidation states estimated via linear calibration to reference compounds (UO₂, UO₃, CaUO₄, UMoO₅, LaUO₄, UTi₂O₆/UTi₂O₇, Ca₃UO₆, USiO₄). μ-EXAFS for U at NSLS-II beamline 4-BM (XFM), k-range 3.0–12.0 Å⁻¹, analysis with Athena/Artemis, FEFF paths, amplitude fixed 0.95, fitting first O-shell distances/coordination. Ce L₃-edge μ-XANES/EXAFS 5525–6025 eV; references CeO₂ (Ce⁴⁺) and CePO₄ (Ce³⁺); k-range 2.0–6.0 Å⁻¹; amplitude fixed 0.9; EXAFS fits to Ce₂Si₂O₇ structural model.

- Phase assemblage: All simulants contained glassy Ca–Al–Si matrix with inclusions of U–Zr–O phases and silicates. Identified crystalline phases included c-(U₁−xZrₓ)O₂, t-(U₁−xZrₓ)O₂, m-(Zr₁−xUₓ)O₂, zircon (ZrSiO₄), anorthite (CaAl₂Si₂O₈), wollastonite (CaSiO₃), SiO₂ (quartz and cristobalite), Fe–Ni metallic particles, and Fe oxides/spinel (FeCr₂O₄, Fe₂O₃, Fe₃O₄). Additional Ce-bearing percleveite, (Ce,Nd)₂Si₂O₇, was identified. - Effect of ZrO₂ addition: Batches with added ZrO₂ (MCCI-2, -3) showed more zircon and anorthite; monoclinic ZrO₂ only present when ZrO₂ was added. ZrO₂ addition promotes zircon formation by stabilising Zr-rich (Zr,U)O₂ relative to (U,Zr)O₂. - U oxidation state and coordination: μ-XANES showed U predominantly U⁴⁺. Estimated oxidation states for crystalline U–Zr–O phases: 4.00 ± 0.10 to 4.22 ± 0.10 (similar to UO₂). U in glass matrix slightly more oxidised: 4.32 ± 0.10 and 4.37 ± 0.10. In Fe-rich regions (MCCI-3), U averaged 3.98 ± 0.10, indicating Fe maintained lower local oxygen potential. EXAFS: in c-(U,Zr)O₂, U–O ~2.32 ± 0.01 Å (8-fold), shorter than UO₂ (2.37 Å); in (Zr,U)SiO₄, two U–O shells at 2.27 ± 0.01 Å (×4) and 2.43 ± 0.01 Å (×4); bond-valence sums ~4.3 and ~4.1 v.u. - Ce distribution and redox: Ce broadly distributed in glass and associated with U–Zr–O crystallites; concentrated near Fe-rich regions and pore interiors. μ-XRD/μ-XRF identified Ce-rich Nd-bearing percleveite (Ce₂Si₂O₇) at interfaces between zircon and glass; more prevalent at higher Ce content (MCCI-3). μ-XANES indicated Ce predominantly Ce³⁺ (e.g., 3.02 ± 0.05 near Fe-rich particles; 3.12 ± 0.05 near U-rich particles). In some Ce hosted within c-(U,Zr)O₂ away from Fe, up to 12 ± 4% Ce⁴⁺ detected by linear combination analysis. - Silicate mineralogy and interfaces: Anorthite commonly enveloped U–Zr–O aggregates, indicating formation at U–Zr–O/silicate melt interfaces; wollastonite more abundant in MCCI-2, typically at edges of fused U–Zr–O aggregates; cristobalite closely associated with zircon and Zr-depleted glass pockets; quartz likely unreacted SiO₂. - Fe-phase redox indicators: Coexistence of Fe–Ni metal and Fe²⁺Cr₂O₄ indicated locally low oxygen potential; presence of Fe₃O₄ and Fe₂O₃ elsewhere indicated higher local oxygen potential. These variations correlate with Ce redox heterogeneity. - Mechanism for percleveite: Formed from reaction of U–Zr–O-depleted Ce–Nd–O melt with silicate melt under reducing conditions, consistent with higher Ce³⁺ solubility in silicate melts.

The study addresses how U and a Pu surrogate (Ce) partition among phases and what oxidation states they adopt in Fukushima-relevant MCCI formed under reducing conditions. Multi-modal μ-focus X-ray methods resolve heterogeneous microstructures and redox states at micrometre scales, showing U remains mostly U⁴⁺ in both crystalline and glassy environments, with slightly higher average oxidation in glasses. Ce predominantly exists as Ce³⁺ and forms a distinct silicate, percleveite (Ce₂Si₂O₇), at interfaces between zircon crystallites and the glass, consistent with reducing synthesis and Ce³⁺’s higher solubility in silicate melts. The identification of varied Fe phases demonstrates local redox heterogeneity; regions retaining metallic Fe–Ni correlate with more reduced conditions that stabilise U⁴⁺ and Ce³⁺, while Fe-poor areas allow minor Ce⁴⁺ in UO₂-based crystals. ZrO₂ addition pathway influences zircon generation (more zircon with added ZrO₂), aligning with thermodynamic expectations for silica-rich concretes and rapid quench pathways (favoring anorthite and wollastonite). These findings inform expectations for real Fukushima MCCI: predominance of U⁴⁺; potential for actinide-bearing silicates at melt interfaces; and strong control of local steel-derived redox buffers on actinide valence and phase partitioning. The micro-analytical approach on gram-scale simulants reproduces key phase assemblages observed in large-scale experiments, supporting its application to sub-mm highly radioactive debris samples when available.

This work synthesised and chemically characterised low-radioactivity simulants of Fukushima MCCI incorporating depleted U and Ce (Pu surrogate), revealing a consistent suite of U–Zr–O phases (c-, t-, m-), zircon, anorthite, wollastonite, Fe–Ni metal and Fe oxides, and a Ce-bearing silicate, percleveite (Ce₂Si₂O₇). U is predominantly U⁴⁺, slightly more oxidised in glass than in crystals, while Ce is mostly Ce³⁺, with minor Ce⁴⁺ detectable in Fe-poor regions within UO₂-based crystals. Fe phase distributions indicate local redox heterogeneity that influences Ce valence and partitioning. ZrO₂ addition promotes zircon formation. The success of μ-focus X-ray multimodal analysis on small, low-activity samples demonstrates a practical pathway for comprehensive phase and speciation analysis of highly radioactive Fukushima debris as it becomes available. Future work should (i) examine true Fukushima MCCI to validate surrogate-based inferences, (ii) explicitly incorporate Pu to quantify differences from Ce surrogates, (iii) broaden fission product chemistry and B₄C/control rod effects, and (iv) couple time–temperature–oxygen potential histories to phase evolution to better predict long-term durability and corrosion behavior.

- Real Fukushima MCCI samples were unavailable; conclusions are based on laboratory simulants under controlled reducing conditions and cooling profiles that approximate, but may not replicate, in-reactor histories. - Ce is an imperfect Pu surrogate; Ce⁴⁺ reduces to Ce³⁺ more readily than Pu⁴⁺, so Pu valence and phase partitioning in real debris may differ. - Highly radioactive fission products and B₄C/control rod materials were omitted, potentially affecting phase equilibria and redox. - Small batch scale and potential compositional heterogeneities may influence local redox and microstructures. - Preferred orientation and overlapping reflections limited unambiguous identification of some minor phases in single-pixel μ-XRD; EDS quantification of trace Ce was challenging. - Affinity of elements for glass vs crystalline phases can be sensitive to exact cooling rates and oxygen partial pressure histories not fully constrained here.