Online program Overview Session: NSS-06

Abstract

Radiation simulations are essential for the design, operation, upgrade, and radiation protection assessment of high-energy physics experiments and other complex accelerator or irradiation facilities. However, their outputs are often difficult to organise, document, retrieve, compare, and share consistently, particularly when many studies, configurations, and users are involved. We present the Analysis platform for Radiation Environment Simulations (ARES), a web-based system developed within the LHCb experiment to support the workflow from registration to analysis of radiation simulation results obtained with the FLUKA particle transport code. ARES provides a database-backed interface for registering radiation estimators together with metadata describing the simulation context, scoring configuration, estimator location, and other relevant information. This structured registration makes results traceable, searchable, reproducible, and easier to reuse. Users can browse and query available estimators through a dedicated search interface and open selected data in a standalone visualisation and analysis service. The platform has been deployed for LHCb radiation studies covering prompt radiation and induced activity use cases, supporting quantities such as fluence, dose, and activation-related estimators. By centralising validated results and providing common tools for access and analysis, ARES improves collaboration between stakeholders involved in simulation production, detector studies, operations, upgrades, and radiation protection. This work demonstrates that a dedicated data-management and analysis layer can make radiation simulations more accessible and more effectively exploitable within a large experimental environment. Future developments include support for additional Monte-Carlo based codes and broader applicability beyond current LHCb use cases.

Abstract

Generation-IV sodium-cooled fast reactors impose severe constraints on neutron flux measurement instruments, particularly due to sustained temperatures that can exceed 550 °C [1]. In this context, optical neutron detectors (ONDs) represent a promising alternative to conventional fission chambers. These devices convert the kinetic energy of fission fragments into optical photons via gas-plasma luminescence and transmit the signal via an optical fiber, allowing electronic components to be placed out of pile, away from radiation. However, the photon collection efficiency of current prototypes is only η ≈ 1.78 × 10^-7, limiting detector sensitivity [2]. This work presents a dedicated Monte Carlo simulation code, written in C, that estimates the impact of a new OND geometry—combining a hemispherical fissile deposit  and a parabolic mirror—on photon collection efficiency, while accounting for the thermal radiation generated at high temperatures. The code comprises three coupled Monte Carlo loops: heavy-ion transport in gas (using polynomial moderation laws from COSICAF), plasma-photon propagation by ray tracing, and thermal wall-radiation modeling at 600 °C. A parametric study identified the negligible impact of the focal length of the parabolic mirror around the geometrically calculated value and quantified the impact of the reflective surfaces, and gas density. The proposed geometry achieves η = 5.4 × 10^-6, a factor of 30 improvement over to current prototypes. These results call for progressive experimental validation, from out-of-reactor thermal characterization to in-reactor testing.

Abstract

Gamma-gamma (γ-γ) coincidence is an advanced radiometric detection technique that has been widely demonstrated to reduce detection limits for many radionuclides in a variety of sample types and applications. The superior selectivity of γ-γ coincidence spectrometry in comparison to traditional single detector analysis, reduces the sensitivity of detection limits to changes in the activity of impurity radionuclides. This is particularly true in the case of impurities that do not emit coincident gamma photons (e.g. ¹³⁷Cs). In this work, a GEANT4 model of a dual high-purity germanium γ-γ coincidence system has been developed and validated against experimental measurements. Additionally, a method to generate modelled time-stamped list mode (T-List) data is presented and demonstrated using long and short-lived radionuclides. By processing the modelled T-List data, the effect on the detection limit for an analyte of interest in the presence of an impurity radionuclide can be studied. These preliminary results demonstrate how the detection limit of ¹³⁴Cs shows no measurable dependence on the activity of ¹³⁷Cs over the simulated range of 0.2-100 kBq. Future work will seek to investigate the influencing factors on detection limits for coincidence spectrometry as well as demonstrate the use of γ-γ coincidence alongside radiochemical separations to accelerate analytical timelines.

Abstract

Quantum imaging with annihilation photons has recently emerged as a promising imaging modality [1, 2]. Positron annihilation photons are correlated particles, carrying the same energy of 511 keV and opposite momenta. The technique relies on coincidence measurements of the annihilation photons' energies and positions, with an object placed in front of one of the detectors. As a result, a ghost image of the object forms in the other detector, and the real image is enhanced by background noise reduction, since the noise lacks the photons' correlations. In this work, we explore the limits of the modality for background noise reduction via Geant4 simulations. The coincidence measurements are simulated for an experimental setup encompassing two segmented scintillating detectors, comprised of 64 Gadolinium Aluminum Gallium Garnet (GAGG) crystals and assembled into an 8x8 square matrix with 3.2 mm pitch [3]. We have found that quantum imaging outperforms classical gamma transmission imaging by retaining the contrast-to-noise ratio (CNR) of noiseless conditions regardless of the background noise intensity. We have also tested the penetrative aspect of annihilation gamma rays, exploring the possibility of detecting attenuation variations in optically dense, high-Z materials.

Abstract

To progress synchrotron-based, spatially fractionated X-ray radiotherapy towards clinical trials, the efficacy of clinically relevant treatment fields must be assessed. In this paper, we present the development and validation of a Geant4 model for human-relevant broad beam (BB) and minibeam radiotherapy (MBRT) treatment fields (up to 7 x 7 cm²) in an experimental hutch located 140m from the synchrotron source at the Imaging and Medical Beamline at the ANSTO Australian Synchrotron. Both broad beam and MBRT treatment fields were modelled for different beam energies and field sizes, with two minibeam fields of different peak spacing investigated. BB and MBRT (peak and valley) depth doses were investigated up to 7cm in a solid water phantom. There was excellent agreement in broad beam percentage depth dose curves (within 4%) as well as the minibeam peak depth dose curves (within 6%), with the model and experimental results agreeing within combined uncertainty for all depths and configurations investigated. There was good agreement in valley dose when comparing EBT-XD Gafchromic film to the model with most results within 10% and within uncertainty. However, there was disagreement between the valley dose measured by the microdiamond detector and the Geant4 model up to 15%. This model will be useful for further MBRT experiments performed at IMBL, with future work validating this model for 3D printed patient volumes.

Abstract

Timepix type pixel detectors enable event-by-event radiation measurements [1], but classification of mixed fields remains challenging due to scarce labeled data and detector dependent effects. We present SimpleMC, an Allpix² [2, 3] based framework for generating realistic synthetic Timepix datasets by combining Monte Carlo transport simulation with detector response modeling and data driven refinement. Photon and electron benchmarks are improved using two complementary approaches: a tuned digitizer and a simplified Gaussian smearing model. These approaches enhance agreement with calibration and spectral measurements. For proton and alpha particles, generative models are used to bridge the sim-to-real gap, including Contrastive Unpaired Translation (CUT) and Conditional Diffusion Denoising Probabilistic Models (CDDPM). The models are trained to map simulated cluster representations to experimental-like responses. CDDPM  outperorms CUT in capturing experimental variability and improving generalization to unseen geometries and irradiation conditions. The framework is evaluated across photons, electrons, and heavy charged particles, showing good agreement in both energy related observables and cluster morphology. Additional experimental datasets from complex radiation environments, including fusion relevant tokamak conditions, are collected for future validation. SimpleMC provides a flexible framework for realistic synthetic data generation, supporting detector characterization and the development of advanced simulation based analysis tools. It will be further developed into a user-friendly toolkit with documentation for the broader radiation detection community.

Abstract

The ClearMind collaboration is developing a next-generation Time-of-Flight PET detector aiming at sub-100 ps coincidence resolving time, high detection efficiency and precise spatial resolution. This presentation discusses the modeling of the ClearMind gamma detector, made of crystal with direct photocathode deposition and microchannel plate (MCP) electron multiplication. The simulation model uses Geant4 software to simulate gamma interactions within the crystal and the propagation of optical photons. Photoelectron generation, multiplication and signal processing are described through a parametrized simulation. This model is crutial for both understanding and optimizing the ClearMind prototype and for development of machine learning-based reconstruction algorithms. To validate the model, we compare simulation results with experimental measurments PbF₂, PbWO₄, BaF₂ and LYSO crystals assembled with the Incom HRPPD MCP-PMT in a conventional configuration. We also explore various factors contributing to the detector's time resolution, including photon collection efficiency, photocathode efficiency, single-photon time resolution and signal pile-up effects. Additionally, we investigated new estimators using deep learning models trained on simulation for time reconstruction and event-by-event associated uncertainty estimation enabling us to select the most precise events. The preliminary results show a time resolution below 100 ps.

Abstract

Fast simulation of carrier generation, transport and signal formation is a powerful tool for the optimization of segmented detector arrays targeted to charged particle identification in nuclear physics experiments. A custom numerical tool for the self-consistent computation of the electron-hole dynamics has been used to study ion tracks in silicon and non-silicon detectors. The code includes the effect of thermal diffusion and Coulomb interaction among the mobile carriers in a microscopic fashion, essential for the correct simulation of high-density ionization tracks in which plasma-like effects (field screening, transient distortions, collection slowdown, and marked nonlinearities) become significant, like in the case of heavy ion interactions. We addressed the challenging case of intermediate mass fragments stopping in the typically few hundred microns of the first detection layer of a telescope configuration, where particle identification critically relies on pulse-shape analysis. Simulations were carried out for several ions - from Li6 to Na22 - impinging on a silicon detector with energies up to the punch-through condition. We studied the impact of the relevant properties of detector and preamplifier response and of the effectiveness of different pulse-shape algorithms to assess the minimum energy threshold and the achievable resolution in charged particle separation in mass (A) and atomic number (Z).