AVS 72 Session AC-ThP: Actinides and Rare Earths Poster Session

Low dimensionality of magnetism and of correlated f-electron physics are current high interest research topics in condensed matter. Here, we compare and contrast the properties of two layered van der Waals uranium alloys with a focus on the diversity of their U 5f electronic structures measured by angle-resolved photoemission (ARPES), and on new f-spectroscopic features not previously reported.

UOTe crystallizes in (Te-U-O-U-Te) quintuplet layers and exhibits bilayer (uddu) antiferromagnetic ordering of local moments below 165K [1]. The U 5f states are found to be well localized at ~1 eV below E_F similar to UO₂, but with an additional weak satellite peak at -0.2 eV of unexplained origin that exhibits a Kondo-like temperature dependence. Moreover, the AFM ordering has a distinct effect on the relative energy alignment of the localized f-states between the zone center and the zone boundary. ARPES of UOTe also exhibits unusual superstructure replicas of the Fermi surface that result from an exotic incommensurate structural modulation of its vdW gap between Te layers.

The UTe₃ crystal structure contains two Te square-net planes that ubiquitously promote charge density wave (CDW) formation in the rare-earth tritellurides. Here strong U 5f character near E_F with strong temperature-dependent Kondo coherence is found in a shallow electron pocket whose high density of states is consistent with its ferromagnetic ordering below 20K [2]. The strong f-scattering dominates the otherwise delicate imperfect Fermi surface nesting of light mass Te p-bands, that results in the suppression of CDW ordering. Finally, an unusual momentum-dependent ‘negative’ spectral weight profile for zone-folded Te p-bands is observed and discussed as an interference phenomena.

[1] C. Broyles et al., Adv. Mater. 37, 2414966 (2025).
[2] J. Shotton et al., arXiv: 2603.03509 (2026).

Accurate identification of fissile isotopes is a central challenge in nuclear forensics and safeguards. Fission Track Analysis (FTA) using solid-state nuclear track detectors (SSNTDs) provides high sensitivity but is limited by manual processing and restricted isotopic discrimination. This work presents an automated, physics-informed framework for fissile isotope characterization combining Monte Carlo simulations, synthetic data generation [1], and image-based analysis.

GEANT4 simulations were performed for ²³⁵U, ²³³U, and ²²⁹Th to model fission fragment transport and cluster formation in LEXAN® SSNTDs with silica aerogel spacers. Synthetic track images were generated to study the effect of aerogel thickness on cluster morphology. Results show that increasing aerogel thickness enhances cluster separation and improves signal-to-noise ratio up to an optimal range, beyond which performance saturates due to pixel-resolution constraints and track truncation effects.

An automated image-processing pipeline (FTA-Finder) was developed to identify fission sites, separate overlapping clusters, and suppress background tracks. The fraction of well-resolved clusters increases with aerogel thickness, indicating improved reconstruction robustness.

Isotopic discrimination was evaluated using reconstructed track-length histograms and double-Gaussian fits for heavy and light fission fragments. A complementary analysis based on Real Flight Path (RFP) distributions showed clear isotope-dependent peak shifts. For 0.1 µm grains embedded in 50 µm aerogel (ρ = 0.033 g/cm³), ²³⁵U exhibited higher RFP peak positions than ²³³U, while ²²⁹Th showed the lowest values due to differences in fission-fragment energy and mass distributions.

Statistical analysis over 30 synthetic realizations showed significant separation between ²³⁵U and ²²⁹Th at ~3σ, while discrimination between ²³⁵U and ²³³U approached the resolution limit (~1σ). Peak deviations remained below 0.69%, confirming stable reconstruction performance.

All results are based on simulation and synthetic data. Future work will focus on experimental validation using neutron irradiation in reactor environments and extension to additional detector configurations. The proposed framework provides a scalable pathway toward quantitative isotopic fingerprinting for nuclear forensic and safeguards applications.

Reference:
[1] R. Babayew, Y. Yehuda–Zada, N. Elgad, J. Lorince, I. Orion, A. Weiss, G. Katarivas Levy, I. Halevy, Simulation tools for improvement of the fission track analysis method for nuclear forensics (2024), JRNC, Akadémiai Kiadó, Budapest, Hungary, DOI: 10.1007/s10967-023-09313-5.

Automated fission-track analysis (FTA) frameworks such as FTA-Finder rely on Monte Carlo–generated synthetic training and validation imagery, typically produced with a single particle-transport code (GEANT4) [1]. A validation strategy built on one simulation engine cannot, by itself, show that the reconstructed forensic observables — fission-center localization, cluster-diameter reconstruction, and isotopic separability — reflect the physical system rather than artifacts of a particular code. This work reports a systematic cross-model assessment of the raw fission-fragment stopping-power and range data underlying FTA-Finder’s synthetic imagery.

Raw range–energy data for a single representative light (⁹⁵Sr, 100 MeV) / heavy (¹³⁸Cs, 67 MeV) ²³⁵U fission-fragment pair — used in place of a full isotope-resolved survey so that the raw-data source remains the only variable under comparison — were generated via five independent routes: GEANT4 Monte Carlo transport; analytical Bethe–Bloch integration; the PDG-recommended combined electronic–nuclear stopping formalism; a Bragg–Kleeman scaling integral; and the SRIM-2013 empirical database, used as reference. Each source was harmonized into a common schema and propagated, unchanged, through the existing FTA-Trainer/FTA-Finder pipeline.

Once a consistent Barkas effective-charge treatment was applied, all five methods agreed with SRIM-2013 to within ~1–9% in projected range across four detector-relevant materials (UO₂ fuel grain, aluminum catcher foil, LEXAN® SSNTD, and silica-aerogel spacer) — vs. 30–160% before correction — showing that agreement is governed by a small number of shared physical parameters rather than code-specific implementation choices.

As a first image-level demonstration, synthetic SSNTD cluster images were generated directly from the PDG-method raw data and processed with the unmodified FTA-Finder pipeline; fission-center localization and cluster-boundary reconstruction were consistent with the GEANT4-based baseline established previously [1] (Supplemental Material).

These results support the conclusion that FTA-Finder is adaptive to, and its forensic conclusions largely independent of, the specific route used to generate its raw training data — an important property for adoption across laboratories with heterogeneous simulation infrastructure. Full image-level robustness metrics for all five sources are in progress.