MDA reduction in HPGe gamma-ray spectrometry using a digital-twin-based intrinsic background construction

Accurate radionuclide determination in radioactive waste via HPGe spectrometry is often limited by background uncertainty, particularly in spectra dominated by Compton continua. Conventional side-window interpolation ties detection limits to counting statistics, resulting in elevated minimum detectable activities (MDAs) and ambiguous peak identification. This study proposes a digital-twin-based framework for explicit construction of intrinsic background spectra with direct statistical uncertainty control. Using a high-fidelity Geant4 Monte Carlo model, background contributions from dominant radionuclides are generated in list-mode with high statistical depth. Intrinsic background spectra for arbitrary activities are reconstructed through virtual-time post-processing of event-level data, avoiding repeated simulations. This framework enables independent suppression of extrinsic and intrinsic background uncertainties without extending physical counting time, thereby lowering detection limits and MDAs. Experimental validation using a certified reference material confirms quantitative accuracy with MDA reductions of approximately 46-75 %, enabling quantification of near-threshold radionuclides. Application to real radioactive waste samples demonstrates two key advantages: ⁹⁴Nb, previously reported as below the MDA, was quantitatively recovered, and background-induced false-positive peaks were objectively rejected based on physically reconstructed continua. These results confirm that radionuclide detectability is governed primarily by background uncertainty rather than magnitude. Digital-twin-based background construction provides a practical route to improved characteristic limits and qualitative reliability in complex radioactive waste characterization.