3D Segmented Water-Based Liquid Scintillator

• 3D segmented water-based liquid scintillator is defined as a detector using water-rich scintillators with explicit three-dimensional position encoding via hardware voxelization, optical self-segmentation, or fast-timing reconstruction.
• Various architectural designs, such as physically segmented cells with reflective separators, opaque systems with light confinement, and monolithic setups with advanced photosensors, balance water fidelity and light yield.
• Performance tests demonstrate improved Cherenkov/scintillation separation, effective localized light collection with low crosstalk, and promising radiation stability, making these detectors ideal for neutrino and low-threshold tracking applications.

A 3D highly-segmented water-based liquid scintillator is a detector class that combines a water-rich scintillating medium with explicitly three-dimensional position encoding at approximately centimeter or sub-centimeter scales. In the literature, the term covers several related but distinct realizations: physically voxelized transparent WbLS detectors with optically isolated cells and orthogonal wavelength-shifting fiber readout; opaque or highly scattering WbLS systems in which the liquid itself confines scintillation light and thereby produces optical pseudo-segmentation; and large monolithic WbLS detectors that are not hardware-segmented but aim to recover 3D topology through fast timing, imaging photosensors, and Cherenkov/scintillation separation. The shared objective is to retain water-target relevance while recovering low-threshold tracking, calorimetry, neutron sensitivity, and event-topology information that are difficult to obtain in pure water detectors (Alonso et al., 2014, Fischer, 2018, Onda et al., 25 Jul 2025, Li et al., 15 Aug 2025, Collaboration, 2024, Che et al., 11 May 2026).

1. Conceptual scope and architectural classes

Water-based liquid scintillator is defined in the large-detector literature as a mixture of pure water and oil-based liquid scintillator made stable by a surfactant that forms micelles, thereby producing a homogeneous mixture (Fischer, 2018). Within that chemistry space, “high segmentation” has acquired three non-equivalent meanings. The strictest meaning is hardware voxelization, in which WbLS is partitioned into optically isolated cubic cells. A second meaning is optical self-segmentation, in which an opaque WbLS confines light close to its origin so that a sparse fiber lattice can reconstruct local topology without physical voxel walls. A third, weaker meaning is reconstruction-defined segmentation in monolithic detectors such as Theia, where fast timing and imaging-capable photosensors provide vertexing, track reconstruction, multiple-ring separation, fiducialization, and event-topology discrimination in a continuous volume rather than in mechanically separated cells (Fischer, 2018, Collaboration, 2024, Onda et al., 25 Jul 2025, Li et al., 15 Aug 2025, Che et al., 11 May 2026).

Architecture	Representative systems	Reported characteristic
Physically segmented transparent WbLS	WbLS tracking detector; 3D segmented WbLS prototype	$1\ \mathrm{cm}^3$ cells, orthogonal WLS fibers, reflective separators (Onda et al., 25 Jul 2025, Li et al., 15 Aug 2025)
Optically self-segmenting opaque WbLS	1-liter oWbLS prototype; pilot 3D-projection oWbLS detector	Light confinement by strong scattering, fiber-lattice readout (Collaboration, 2024, Che et al., 11 May 2026)
Monolithic 3D optical imaging	Theia; ASDC	No physical segmentation; topology reconstructed from timing and optical information (Fischer, 2018, Alonso et al., 2014)

The earliest broad WbLS concept papers were monolithic. The Advanced Scintillator Detector Concept proposed a 30–100 kiloton-scale WbLS detector with high-efficiency photon detection, ultra-fast timing photosensors, and isotope loading, but it also explicitly pointed to the Raghavan Optical Lattice or LENS-style cubical lattice as a geometry that could “cleanly segment” the detector into many well-resolved independent cells, at the cost that liquid circulation would not be possible (Alonso et al., 2014). This historical split between monolithic optical imaging and true hardware segmentation continues to structure the field.

2. Active media and optical regimes

The chemistry of highly segmented WbLS detectors spans several optical regimes rather than a single canonical formulation. In early transparent PC-based WbLS, two mixtures were characterized in detail: WbLS-1 with $0.4\%$ scintillator by mass and WbLS-2 with $0.99\%$ scintillator by mass. Their fitted scintillation light yields were $19.9 \pm 1.1 \pm 2.0$ photons/MeV and $108.9 \pm 0.8 \pm 10.9$ photons/MeV, respectively, with quenching parameters $0.70 \pm 0.12 \pm 0.07\ \mathrm{mm/MeV}$ and $0.44 \pm 0.01 \pm 0.04\ \mathrm{mm/MeV}$, substantially larger than the $0.07 \pm 0.01 \pm 0.01\ \mathrm{mm/MeV}$ reported for the pure LS reference (Bignell et al., 2015). A later Brookhaven 1-ton study of Gd-compatible WbLS measured intrinsic light yield as a function of concentration from $0.35\%$ to $1.0\%$ by mass, obtaining $0.4\%$0 ph/MeV at $0.4\%$1 and $0.4\%$2 ph/MeV at $0.4\%$3, with $0.4\%$4 identified as the minimum concentration for chemical stability (Gwon et al., 17 Dec 2025).

A different transparent regime is represented by the Triton X-100 formulation, which uses 86 vol% distilled water, 13 vol% Triton X-100, 1 vol% LAB, approximately $0.4\%$5 PPO in the final WbLS, and $0.4\%$6 vitamin C. For this material, the attenuation lengths were measured as $0.4\%$7 at $0.4\%$8 and $0.4\%$9 at $0.99\%$0, the viscosity as $0.99\%$1, the micelle-size peak as $0.99\%$2 with RMS $0.99\%$3, and the $0.99\%$4-interaction light yield as $0.99\%$5 photons/MeV. Its fluorescence timing under UV excitation was dominated by a $0.99\%$6 component with $0.99\%$7 weight, and its neutron/gamma pulse-shape discrimination was reported as comparable to LAB + $0.99\%$8 PPO (Steiger et al., 2024).

At much higher scintillating fraction, the segmented $0.99\%$9 WbLS prototype used a 90% water by mass and 10% LAB-based active component formulation with $19.9 \pm 1.1 \pm 2.0$0 PPO and $19.9 \pm 1.1 \pm 2.0$1 MSB. From the nominal LS primary yield of about 9,000 to 12,000 photons/MeV, the expected WbLS primary yield was estimated as about 900–1200 photons/MeV, and the active detector structure reached 81% water by mass excluding the outer box (Li et al., 15 Aug 2025).

Opaque WbLS represents a distinct optical regime. In the 1-liter, 32-channel prototype, the optimized reduced scattering length for oWbLS2 was $19.9 \pm 1.1 \pm 2.0$2, with effective fitted absorption length $19.9 \pm 1.1 \pm 2.0$3 and light yield about $19.9 \pm 1.1 \pm 2.0$4 photons/MeV (Collaboration, 2024). In the pilot 3D-projection detector, beam data showed tighter confinement than a Geant4 model with a $19.9 \pm 1.1 \pm 2.0$5 scattering length, placing the effective scattering length well below $19.9 \pm 1.1 \pm 2.0$6 and supporting the intended strongly scattering regime (Che et al., 11 May 2026).

This distribution of formulations suggests that “3D highly-segmented WbLS” is not tied to one optical strategy. Transparent low-loading mixtures emphasize coexistence of Cherenkov and scintillation channels; water-rich 10% active-component mixtures emphasize water-target fidelity with moderate light yield; opaque mixtures trade long transport for local light confinement. A plausible implication is that segmentation geometry and WbLS chemistry are co-designed rather than sequentially chosen.

3. Segmentation hardware and readout geometries

The most direct hardware realization is the WbLS tracking detector developed for neutrino interaction measurements on a water target. Its basic unit is a $19.9 \pm 1.1 \pm 2.0$7 cubic cell of WbLS, optically segmented by reflective separators and read out through $19.9 \pm 1.1 \pm 2.0$8 Kuraray Y11(200)M fibers in three orthogonal directions. The beam-test prototype had five layers of $19.9 \pm 1.1 \pm 2.0$9 cells, for a total active volume of $108.9 \pm 0.8 \pm 10.9$0, 80 cells, and 56 readout channels; only the upstream layer had all three views. The baseline beam-test WbLS was approximately 70% water, 20% surfactant, and 10% liquid scintillator. After the beam test, an improved U2 sample with 65% water and 20% liquid scintillator, enabled by IGEPAL CO-630, increased the measured light yield by $108.9 \pm 0.8 \pm 10.9$1 relative to the S5/S1-like baseline, while PTFE separator material with 93% reflectivity produced an additional $108.9 \pm 0.8 \pm 10.9$2 gain (Onda et al., 25 Jul 2025).

The later $108.9 \pm 0.8 \pm 10.9$3 3D segmented WbLS prototype used a different engineering solution. Each $108.9 \pm 0.8 \pm 10.9$4 voxel was enclosed by a lightweight separator built from a 1.2 mm Divinycell H80 foam core sandwiched between two 3M DF2000MA reflector films, for total separator thickness 1.4 mm. The prototype contained 27 cubes arranged in three layers of $108.9 \pm 0.8 \pm 10.9$5, read out by 18 Kuraray Y11 double-cladding fibers coupled to Hamamatsu S13360-1350CS SiPMs. The architecture presently provides two orthogonal projections, while the paper explicitly notes that a vertical WLS fiber not included in this work could be added later to provide a third projection (Li et al., 15 Aug 2025).

Opaque WbLS developed a different route to effective segmentation. The 1-liter detector used a $108.9 \pm 0.8 \pm 10.9$6-side cube with 16 WLS fibers in each of two vertical planes, one $108.9 \pm 0.8 \pm 10.9$7 and one $108.9 \pm 0.8 \pm 10.9$8, offset in $108.9 \pm 0.8 \pm 10.9$9, for 32 channels total. Fiber pitch was $0.70 \pm 0.12 \pm 0.07\ \mathrm{mm/MeV}$0 in $0.70 \pm 0.12 \pm 0.07\ \mathrm{mm/MeV}$1 and $0.70 \pm 0.12 \pm 0.07\ \mathrm{mm/MeV}$2, and $0.70 \pm 0.12 \pm 0.07\ \mathrm{mm/MeV}$3 in $0.70 \pm 0.12 \pm 0.07\ \mathrm{mm/MeV}$4. The architecture was explicitly presented as an alternative to both external imaging and physical voxelization, because light is stochastically confined about its origin and only nearby fibers receive substantial signal (Collaboration, 2024). The pilot 3D-projection oWbLS detector extended that idea to a larger continuous volume of $0.70 \pm 0.12 \pm 0.07\ \mathrm{mm/MeV}$5 instrumented with three orthogonal fiber planes on a 10 mm pitch, for 320 total Kuraray Y11 fibers and a nominal effective voxel of $0.70 \pm 0.12 \pm 0.07\ \mathrm{mm/MeV}$6 without hard voxel walls (Che et al., 11 May 2026).

By contrast, Theia remains a monolithic $0.70 \pm 0.12 \pm 0.07\ \mathrm{mm/MeV}$7-kiloton target viewed by more than 100,000 photosensors with effective photocoverage greater than 90%. It is therefore not highly segmented in the detector-engineering sense. Its relevance is that fast-timing photosensors such as LAPPDs, together with WbLS, are expected to provide vertex reconstruction, charged-particle track reconstruction, ring imaging, multiple-ring separation, and event-topology reconstruction in a continuous medium. In that restricted sense, it realizes effective 3D granularity through reconstruction rather than through cells (Fischer, 2018).

4. Reconstruction methods and measured performance

The performance envelope of 3D highly-segmented WbLS depends on which optical regime is chosen. In transparent low-loading WbLS, fast timing can isolate a prompt photon sample that is majority Cherenkov. Using an LAPPD and a conventional PMT with effective $0.70 \pm 0.12 \pm 0.07\ \mathrm{mm/MeV}$8 resolution, three WbLS mixtures loaded at 1%, 5%, and 10% scintillator were characterized. The optimized prompt windows were 300 ps for 1% WbLS and 200 ps for both 5% and 10%, yielding Cherenkov purities of 80.4%, 68.6%, and 64.3%, respectively. The corresponding fitted scintillation rise times were $0.70 \pm 0.12 \pm 0.07\ \mathrm{mm/MeV}$9, $0.44 \pm 0.01 \pm 0.04\ \mathrm{mm/MeV}$0, and $0.44 \pm 0.01 \pm 0.04\ \mathrm{mm/MeV}$1, with dominant fast decay constants around 2.2–2.4 ns. This demonstrates that low-scintillator-fraction WbLS combined with LAPPD-class timing preserves a usable prompt directional component, although the available timing leverage is concentrated within only a few hundred picoseconds (Kaptanoglu et al., 2021).

Transparent physically segmented prototypes have so far been limited primarily by light collection. In the 80-cell WbLS tracker, the detector requirement for $0.44 \pm 0.01 \pm 0.04\ \mathrm{mm/MeV}$2 MIP detection at $0.44 \pm 0.01 \pm 0.04\ \mathrm{mm/MeV}$3 photoelectrons threshold was $0.44 \pm 0.01 \pm 0.04\ \mathrm{mm/MeV}$4 per fiber readout. The beam-test prototype reached a maximum of $0.44 \pm 0.01 \pm 0.04\ \mathrm{mm/MeV}$5 per fiber readout, with typical measured light yield less than 4 p.e. per fiber and mean cross-talk roughly 5%, generally less than 10% (Onda et al., 25 Jul 2025). In the later $0.44 \pm 0.01 \pm 0.04\ \mathrm{mm/MeV}$6 segmented WbLS prototype, the measured WbLS light yield for cosmic MIPs was 5.4 p.e./channel/MIP as most probable value and 6.0 p.e./channel/MIP on average, compared with 39.1 and 42.5 p.e./channel/MIP for the geometrically identical pure-LS reference. Optical crosstalk was 2.29% for WbLS and 1.90% for LS. The WbLS signal was judged sufficient for tracking, and stopping protons were expected to produce up to five times more light near the Bragg peak (Li et al., 15 Aug 2025).

Opaque WbLS has already demonstrated stronger localization. In the 1-liter, 32-channel prototype, a center-of-mass-like reconstruction calibrated with $0.44 \pm 0.01 \pm 0.04\ \mathrm{mm/MeV}$7 simulated events gave a mean position reconstruction error of 4.4 mm for 1.6 MeV-equivalent events and 7.4 mm for 0.8 MeV-equivalent events, with energy resolutions of $0.44 \pm 0.01 \pm 0.04\ \mathrm{mm/MeV}$8 and $0.44 \pm 0.01 \pm 0.04\ \mathrm{mm/MeV}$9, respectively (Collaboration, 2024). In the pilot 3D-projection detector, charge from matched views was assigned to voxels through

$0.07 \pm 0.01 \pm 0.01\ \mathrm{mm/MeV}$0

and track-like events were characterized with a charge-weighted PCA. For 500 MeV protons, about 94% of the normalized charge lay within $0.07 \pm 0.01 \pm 0.01\ \mathrm{mm/MeV}$1, about 5% within $0.07 \pm 0.01 \pm 0.01\ \mathrm{mm/MeV}$2, and charge beyond $0.07 \pm 0.01 \pm 0.01\ \mathrm{mm/MeV}$3 was consistent with zero, establishing very strong transverse confinement. The same detector yielded a single-channel timing resolution of $0.07 \pm 0.01 \pm 0.01\ \mathrm{mm/MeV}$4–$0.07 \pm 0.01 \pm 0.01\ \mathrm{mm/MeV}$5 with good photostatistics, per-voxel timing around 0.16 ns, and half-track timing around 0.05 ns (Che et al., 11 May 2026).

Material-side discrimination channels can also be relevant. The Triton X-100 WbLS measured a maximal tail-to-total difference $0.07 \pm 0.01 \pm 0.01\ \mathrm{mm/MeV}$6, with optimal tail start $0.07 \pm 0.01 \pm 0.01\ \mathrm{mm/MeV}$7 after pulse peak, which the authors describe as comparable to LAB + $0.07 \pm 0.01 \pm 0.01\ \mathrm{mm/MeV}$8 PPO (Steiger et al., 2024). This suggests that segmentation and PSD are not mutually exclusive, although cell-scale PSD has not yet been demonstrated.

5. Radiation damage, mixing, and operational stability

Radiation damage has been directly measured for a 5% WbLS exposed to 201 MeV proton beams. No damage to scintillation light yield was evident for doses up to $0.07 \pm 0.01 \pm 0.01\ \mathrm{mm/MeV}$9. At $0.35\%$0, the WbLS light-yield loss was $0.35\%$1; at $0.35\%$2, it was $0.35\%$3. Increased optical absorption accounted for only part of that loss: the optical contribution at $0.35\%$4 was $0.35\%$5, implying an additional intrinsic scintillation-yield degradation. The paper parameterized the attenuation contribution as

$0.35\%$6

with $0.35\%$7 the mean optical path length and $0.35\%$8 the effective damage-induced absorbance. No measurable recovery was observed over about one week, and a conservative bulk-phantom extrapolation gave approximately $0.35\%$9 annual light-yield reduction after a year of clinical operation (Bignell et al., 2015).

For highly segmented detectors, this suggests two opposite effects. Short voxel-scale optical paths should suppress the attenuation-related part of damage because $1.0\%$0 is smaller. By contrast, intrinsic scintillation-yield loss remains local, so repeatedly irradiated cells or poorly mixed volumes could accumulate percent-level gain drifts on the hundreds-of-gray scale. That specific concern is not resolved experimentally in current segmented prototypes and remains architecture-dependent (Bignell et al., 2015).

Bulk mixing and scale-up studies provide the operational background for any future segmented system. In the Brookhaven 1-ton WbLS detector, a 1% DIN/PPO WbLS prepared in situ reached apparently uniform dispersion in about 20 minutes despite a full circulation time of approximately 50 minutes, and the same program reported scattering length higher than 30 meters at 450 nm (Xiang et al., 2024). The 30-ton Brookhaven demonstrator, although monolithic and not segmented, established staged injection at 0.3%, 0.75%, and 1% WbLS by mass, continuous circulation and purification, direct contact with passivated 316L stainless steel, and two-month PMT soak tests in WbLS without statistically significant degradation (Andrade et al., 20 Mar 2026). These studies do not solve segmented-detector fluid management, but they do show that ton-scale WbLS handling, purification, and monitoring are already technically mature enough to support more complex geometries.

6. Physics use cases, unresolved issues, and development trajectory

The principal application driver is neutrino detection on a water-relevant target with more complete final-state reconstruction than a water Cherenkov detector can provide. The $1.0\%$1 segmented WbLS near-detector concept was motivated by the need for few-percent-level neutrino-nucleus interaction constraints for water-based far detectors such as Hyper-Kamiokande, and specifically by the need to reconstruct low-energy hadrons and protons below $1.0\%$2 in a target that still retains more than 80% water by mass (Li et al., 15 Aug 2025). The earlier WbLS tracking detector was similarly framed as a means to characterize neutrino interactions on a water target while recovering three-dimensional tracking analogous to SuperFGD (Onda et al., 25 Jul 2025).

At larger scale, Theia and ASDC show that WbLS is also being developed for a broad program including long-baseline oscillations, solar, reactor, and supernova neutrinos, neutrinoless double beta decay, and proton decay, with fast optical reconstruction rather than hardware segmentation as the main organizing principle (Fischer, 2018, Alonso et al., 2014). Opaque WbLS prototypes extend the application space further toward point-like topology reconstruction, spectroscopy, and more general localized-radiation detection (Collaboration, 2024, Che et al., 11 May 2026).

Several limitations remain structurally important. Physically segmented transparent WbLS prototypes are still below their nominal light-yield targets, and separator reflectivity has emerged as a first-order performance determinant (Onda et al., 25 Jul 2025). The $1.0\%$3 water-rich prototype has only two projections at present, with a third vertical view explicitly left for future implementation (Li et al., 15 Aug 2025). Opaque WbLS has demonstrated point-like localization and beam-track imaging, but not yet full multi-site separation thresholds in a large detector (Collaboration, 2024, Che et al., 11 May 2026). Monolithic WbLS R&D establishes medium handling, stability, and Cherenkov/scintillation coexistence, but not cell-to-cell calibration, voxel-level drift, or segmented fluid management (Xiang et al., 2024, Andrade et al., 20 Mar 2026).

Chemistry diversification is also ongoing. The p-dioxane/tellurium/water scintillator is surfactant-free and water-compatible rather than a conventional water-dominant WbLS, and it is motivated by tellurium loading rather than segmentation (Liang et al., 20 May 2025). Water-based quantum-dot scintillator provides another aqueous route, with CdS/ZnS QDs phase-transferred into water, preserved blue emission, inferred scintillation yield around 4000 photons/MeV, and few-nanosecond timing, but without any segmented-detector validation (Zhao et al., 2024). These adjacent formulations indicate that the broader design space now extends beyond classical surfactant-stabilized LAB-in-water emulsions.

Taken together, current work supports a precise characterization of the field. A 3D highly-segmented water-based liquid scintillator is no longer only a conceptual extrapolation from monolithic WbLS. It now includes true $1.0\%$4 voxelized water-rich detectors, optically self-segmenting opaque WbLS systems with millimeter-to-centimeter localization, and monolithic fast-timing WbLS detectors that achieve segmentation-like behavior in reconstruction. What remains unsettled is not the existence of the concept, but the optimal compromise among water fraction, light yield, separator technology, Cherenkov purity, local calibration stability, and scalable readout architecture.