image: This review summarizes principles, materials, and pulse shape discrimination (PSD) for neutron detection. Advanced organic, crystal, and glass scintillators are evaluated for thermal and fast neutron measurements. Photodetectors, digitization, and AI algorithms are integrated to enhance neutron–γ PSD. Future perspectives cover scalable manufacturing, radiation hardness, and real-time mixed-field deployment.
Credit: Sidorenkov A., Qian S., Hu P., et al., (2026). Advanced scintillators for neutron-gamma detection. The Innovation Materials 4(4), 100245.
A neutron passing through matter is invisible in the most literal sense: it carries no charge, ionizes almost nothing, and leaves no track for electronics to follow. Detectors therefore convert neutrons into light — and then face a harder problem, because the gamma rays that fill the same environment produce light too.
A review published in The Innovation Materials, by 17 authors from Chinese and Russian institutes led by Sen Qian of the Institute of High Energy Physics, maps how that separation is achieved today, from nuclear physics to signal processing.
The physics sets the options. Fast neutrons are detected indirectly, through nuclei they knock forward, whose recoil excites the scintillator. Thermal neutrons are captured instead, by helium-3, boron-10, lithium-6 or gadolinium-155 and -157, and the capture products deposit energy locally. Each route leaves a different pulse shape, and pulse shape discrimination — comparing, say, the fast and slow parts of a signal — is how detectors decide which particle arrived.
Materials determine how well that works. The review sets out the trade-offs across liquid scintillators such as EJ-301, lithium-loaded plastics, halide crystals (CLYC, CLLB, NaI:Tl,Li), and lithium-6 or gadolinium-doped glasses. Light output spans a factor of nearly twenty: about 8,600 photons per MeV for EJ-254 plastic, roughly 70,000 for Cs2LiYCl6:Ce. The discrimination figure of merit (FOM), the standard measure of how cleanly neutron and gamma populations separate, ranges from about 1.0 to 4.2 across the same materials, and depends as much on crystal quality, photodetector and energy threshold as on composition.
That dependence is the field's weak point. Conventional charge-comparison discrimination degrades sharply below roughly 200 keV of electron-equivalent energy, where the two pulse populations overlap — precisely the range that matters for weak sources and shielded material.
The review's most forward-looking section is on machine learning. Neural networks operating directly on sampled waveforms, hybrid models combining convolutional layers with hand-built features, and specialized architectures for pulse pile-up recover separation where classical methods fail, though the reported gains at low energy are modest — one pulse-coupled neural network improved FOM from 1.04 to 1.10 — and the authors note that training data and generalization across detector systems remain open problems. Digital pulse shaping, by comparison, lifts the CLYC FOM by about 14% without any learning.
No new measurements are reported; the conclusions rest on published performance under differing conditions, which limits direct comparison between materials. Scalable manufacturing, radiation hardness, and real-time operation in mixed fields are identified as the requirements that will decide which of these materials reaches portal monitors, reactors, medical imaging and space instruments.
###
Citation: Sidorenkov A., Qian S., Hu P., et al., (2026). Advanced scintillators for neutron-gamma detection. The Innovation Materials 4(4), 100245. https://doi.org/10.59717/j.xinn-mater.2026.100245
Journal
The Innovation Materials
Method of Research
Literature review
Subject of Research
Not applicable
Article Title
Advanced scintillators for neutron-gamma detection
Article Publication Date
20-Sep-2026
COI Statement
The authors declare no competing interests.