ESSnuSB Far Detector

• ESSnuSB Far Detector is a megaton-class underground water-Cherenkov system located at the Zinkgruvan mine, optimized for measuring the leptonic Dirac phase via νμ→νe appearance at the second oscillation maximum.
• Its design varies across studies—from a single 538 kt tank to multi-module configurations (two 270 kt or four 135 kt modules) with high-efficiency PMT instrumentation and rigorous background suppression.
• The detector also supports atmospheric, solar, and supernova neutrino research using advanced timing, reconstruction techniques, and systematic controls for precise event identification.

Searching arXiv for recent ESSnuSB far-detector papers to ground the article. Attempting arXiv search for "ESSnuSB far detector". The ESSnuSB far detector denotes the large underground water-Cherenkov detector complex proposed for the ESSnuSB long-baseline neutrino program, sited at the Zinkgruvan mine in Sweden about 360 km from the ESS linac in Lund and optimized for measurements at the second oscillation maximum. Its primary role is the precision determination of the leptonic Dirac phase $\delta_{\rm CP}$ through $\nu_\mu \rightarrow \nu_e$ appearance, but the same detector volume is also used in the literature for atmospheric, solar, and core-collapse supernova neutrino studies. Across the current arXiv literature, the detector appears in closely related but not fully identical design conventions: a 538 kt single-tank reference in recent proceedings and supernova studies, a 540 kt two-module system in the design and atmospheric program, and a 540 kt four-module baseline in one ESSnuSBplus proceedings contribution (Ghosh, 16 Jun 2026, ESSnuSB et al., 25 Jun 2026, Collaboration et al., 2023, ESSnuSB et al., 2024, ESSnuSB et al., 15 Jan 2026).

1. Configuration, mass conventions, and site

The far detector is consistently located at the Zinkgruvan mine, with a baseline of about $L = 360$ km from the ESS source in Lund. The active medium is ultra-pure water, and the detector technology is a classical water-Cherenkov system instrumented with photomultiplier tubes to record Cherenkov light. What varies across publications is the geometrical decomposition and the quoted fiducial mass (Ghosh, 16 Jun 2026, Collaboration et al., 2023, ESSnuSB et al., 25 Aug 2025, ESSnuSB et al., 15 Jan 2026).

Source	Stated configuration	Stated mass
(Ghosh, 16 Jun 2026)	Single ultra-pure water Cherenkov tank	538 kt fiducial mass
(Collaboration et al., 2023, ESSnuSB et al., 2024, ESSnuSB et al., 25 Aug 2025)	Two identical cylindrical Water Cherenkov detectors	540 kt total fiducial mass
(ESSnuSB et al., 15 Jan 2026)	Four identical cylindrical modules in the baseline	540 kt total fiducial mass

The two-module design studies describe two identical cylindrical underground caverns, with each module carrying a fiducial mass of 270 kt after a 2 m fiducial cut from the walls. One review gives each cavern as a steel PMT-frame cylinder of 78 m diameter and 78 m height, with total water volume $\sim 740$ kt per detector and a 2 m cut yielding 270 kt fiducial mass; another atmospheric study uses outer dimensions of 76 m diameter and 76 m height, with a 70 m $\times$ 70 m fiducial cylinder (Collaboration et al., 2023, ESSnuSB et al., 2024). By contrast, one ESSnuSBplus proceedings contribution states a baseline of four identical cylindrical modules of approximately 135 kt each, each about 60 m in diameter and 60 m in height (ESSnuSB et al., 15 Jan 2026). A status summary gives two tanks of roughly 60 m diameter and 60 m height, with total water mass $2 \times 270$ kt and fiducial mass $\simeq 2 \times 250$ kt (Fanourakis et al., 21 Jan 2025).

The underground setting is also described with slightly different levels of specificity. Several design and atmospheric studies state approximately 1 km overburden, or $O(1\,{\rm km\,w.e.})$, and emphasize strong suppression of cosmic-ray muons; one proceedings article, however, simply states “the Zinkgruvan mine” and gives no numeric overburden (Collaboration et al., 2023, ESSnuSB et al., 2024, ESSnuSB et al., 25 Aug 2025, Fanourakis et al., 21 Jan 2025, Ghosh, 16 Jun 2026). This suggests an evolution in design conventions and in the level of engineering detail adopted by different simulation studies, rather than a change in the basic experimental concept.

2. Water-Cherenkov technology and detector instrumentation

The common technological core is a large ultra-pure water-Cherenkov detector using ring imaging to classify charged-current events as electron-like or muon-like. The water is described in one ESSnuSBplus proceedings paper as ultra-pure H$_2$O, deionized, low-radon, with $\le 10^{-14}$ g/g U/Th (ESSnuSB et al., 15 Jan 2026). The design studies identify the far detector explicitly with the Super-Kamiokande/Hyper-Kamiokande lineage, and one performance paper states that the detector closely follows the Hyper-K design, with WCSim GEANT4-based simulation and FitQun reconstruction (Ghosh, 16 Jun 2026, Alekou et al., 2021).

The photodetector layout is again study-dependent in detail but not in principle. The design review and the atmospheric-neutrino study specify inward-pointing 20-inch PMTs covering 30% of the inner surface and outward-pointing PMTs for vetoing entering muons; one review further specifies outward-facing 8-inch PMTs in the veto region (Collaboration et al., 2023, ESSnuSB et al., 2024). A status report instead summarizes the detector as using large-area PMTs covering $\nu_\mu \rightarrow \nu_e$0 of the inner surface, with typical 20-inch PMTs and quantum efficiency around 25% at 400 nm; a separate proceedings note states that each 135 kt module is lined with about 40,000 high-quantum-efficiency 20-inch PMTs providing at least 40% photocathode coverage (Fanourakis et al., 21 Jan 2025, ESSnuSB et al., 15 Jan 2026). Optical parameters quoted in the status report include water attenuation length $\nu_\mu \rightarrow \nu_e$1 m at 400 nm, refractive index $\nu_\mu \rightarrow \nu_e$2, and scattering length $\nu_\mu \rightarrow \nu_e$3 m (Fanourakis et al., 21 Jan 2025).

Timing is an important part of the far-detector concept. The ESS beam is described as operating with about 1.2 $\nu_\mu \rightarrow \nu_e$4s pulses, and one proceedings paper states that PMT signals are digitized with 0.1 ns precision, enabling time-of-flight tagging of the 1.2 $\nu_\mu \rightarrow \nu_e$5s pulses and rejection of out-of-time backgrounds (Ghosh, 16 Jun 2026, ESSnuSB et al., 15 Jan 2026). In the beam-oscillation analysis, this timing enters directly into background control through cuts synchronized to the spill.

3. Event reconstruction, efficiencies, and background suppression

The far detector is optimized for sub-GeV charged-current interactions, especially in the few-hundred-MeV region relevant to the second oscillation maximum. The detection strategy is based on Cherenkov ring-shape identification, topology cuts, and beam timing. The beam backgrounds explicitly considered in one proceedings summary are beam-intrinsic $\nu_\mu \rightarrow \nu_e$6, misidentified $\nu_\mu \rightarrow \nu_e$7 or $\nu_\mu \rightarrow \nu_e$8 events, and neutral-current $\nu_\mu \rightarrow \nu_e$9 events, with a standard Cherenkov-ring-shape identification strategy plus timing cuts matched to the $L = 360$0 beam spill (Ghosh, 16 Jun 2026).

Performance numbers depend on the simulation chain and on the benchmark definition. In the beam-program proceedings, the $L = 360$1-appearance acceptance efficiency reaches approximately 85% around the beam peak, $L = 360$2–$L = 360$3 GeV, while the background rejection efficiency exceeds 99% in the same window (Ghosh, 16 Jun 2026). In the design-study review, full GEANT4-based simulation yields $L = 360$4 charged-current $L = 360$5-like identification efficiencies of about 65% at 0.25 GeV rising to about 80–85% at 0.5 GeV, and $L = 360$6 charged-current $L = 360$7-like identification above 90% for $L = 360$8 GeV; the same source quotes typical single-ring energy resolution $L = 360$9–10% and angular resolution $\sim 740$0 for $\sim 740$1–0.6 GeV (Collaboration et al., 2023). A status report gives a benchmark $\sim 740$2 CCQE efficiency

$\sim 740$3

so that $\sim 740$4 at $\sim 740$5 GeV, together with

$\sim 740$6

and neutral-current rejection $\sim 740$7 (Fanourakis et al., 21 Jan 2025).

The updated 2021 performance study implemented signal and background channels in GLoBES with energy-dependent efficiencies and migration matrices derived from a GEANT4-based simulation tuned to ESSnuSB. There, the updated $\sim 740$8 appearance signal efficiency rises from about 20% at 0.2 GeV to about 60% at 0.4 GeV and then falls gently to about 40% at 0.8 GeV, while $\sim 740$9 disappearance efficiencies are about 80–90% over 0.2–1.0 GeV (Ghosh, 2021). Another 2021 update describes a 538 kt water-Cherenkov detector with 40% PMT coverage, FitQun reconstruction, and WCSim simulation, with charged-current events selected at greater than 90% efficiency in the 0.2–1.0 GeV range through dedicated low-energy optimization (Alekou et al., 2021).

These differing efficiency numbers should not be read as direct contradictions. They correspond to different channels, cuts, fiducial definitions, and benchmark implementations. What remains stable across the literature is the qualitative operating point: high charged-current efficiency in the sub-GeV regime, strong $\times$0 rejection, and decisive use of timing and ring topology for background suppression.

4. Oscillation measurements at the second maximum

The detector’s defining physics role is to measure $\times$1 at the second oscillation maximum. A leading-order appearance expression quoted in the proceedings literature is

$\times$2

with the omitted terms proportional to $\times$3, the matter effect, $\times$4, and related subleading contributions (Ghosh, 16 Jun 2026). More expanded formulae appear in the design and ESSnuSBplus proceedings papers, all emphasizing that the CP-odd contribution is enhanced when the experiment is tuned to the second maximum (Collaboration et al., 2023, ESSnuSB et al., 15 Jan 2026).

The rationale is stated in multiple ways across the literature. The design-study review reports that the genuine CP-violating asymmetry in $\times$5 is roughly $\times$6 larger at the second maximum than at the first, with matter-induced distortions remaining minimal (Collaboration et al., 2023). The status report phrases the gain as an increase of the CP-sensitive interference term by about $\times$7 at $\times$8 km and $\times$9 GeV, where $2 \times 270$0 (Fanourakis et al., 21 Jan 2025). One ESSnuSBplus proceedings paper states that the choice of baseline and spectrum gives roughly a three-fold gain in CP sensitivity (ESSnuSB et al., 15 Jan 2026). Figure-based descriptions in the beam proceedings further note that the 360 km option covers part of the first maximum around 0.4 GeV and the second maximum around 0.15 GeV, with the flux$2 \times 270$1cross-section histogram peaking near 0.2–0.3 GeV (Ghosh, 16 Jun 2026).

Projected physics reach is correspondingly strong but simulation-dependent. Under conservative normalization systematics of 5% for signal and 10% for background, the updated physics-performance study finds CP-violation discovery sensitivities of 10$2 \times 270$2 at 540 km and 13$2 \times 270$3 at 360 km for $2 \times 270$4, with about 70% coverage of the $2 \times 270$5 range above 5$2 \times 270$6 after a 10-year run (Alekou et al., 2021). The status paper reports that after 10 years, about 72% of the possible CP-violating phase range is covered with 5$2 \times 270$7 C.L. to reject the no-CP-violation hypothesis, with expected precision better than $2 \times 270$8 for all allowed values; the same summary quotes total $2 \times 270$9 signal events of about 1600 in $\simeq 2 \times 250$0 mode and 1000 in $\simeq 2 \times 250$1 mode, against backgrounds of about 200 and 150 respectively (Fanourakis et al., 21 Jan 2025). The design-study review quotes CP-violation discovery exceeding 5$\simeq 2 \times 250$2 for $\simeq 2 \times 250$3 of true $\simeq 2 \times 250$4 values under 5% normalization uncertainty, and $\simeq 2 \times 250$5 precision below $\simeq 2 \times 250$6 for all $\simeq 2 \times 250$7 (Collaboration et al., 2023). A more recent beam proceedings figure summary states that at maximal CP violation the far detector reaches about 12$\simeq 2 \times 250$8 after the nominal 10-year run, with worst-case $\simeq 2 \times 250$9 precision $O(1\,{\rm km\,w.e.})$0 (Ghosh, 16 Jun 2026).

The far detector also underpins disappearance and event-rate measurements. One ESSnuSBplus proceedings paper quotes an unoscillated $O(1\,{\rm km\,w.e.})$1 flux at 360 km peaking near $O(1\,{\rm km\,w.e.})$2 GeV with full-width half-maximum of approximately 0.2 GeV, and expects about $O(1\,{\rm km\,w.e.})$3 $O(1\,{\rm km\,w.e.})$4 charged-current events before oscillations over the nominal exposure (ESSnuSB et al., 15 Jan 2026). The design-study review gives example $O(1\,{\rm km\,w.e.})$5-mode appearance yields over the full run: $O(1\,{\rm km\,w.e.})$6 $O(1\,{\rm km\,w.e.})$7-appearance events for $O(1\,{\rm km\,w.e.})$8, $O(1\,{\rm km\,w.e.})$9 for $_2$0, and $_2$1 for $_2$2, against total backgrounds of $_2$3–$_2$4 (Collaboration et al., 2023).

5. Atmospheric-neutrino program and non-standard interactions

Because of its very large water volume, the far detector is also treated in the literature as a 5.4 Mt$_2$5year atmospheric-neutrino observatory over 10 years of operation. The atmospheric analyses assume separate $_2$6-like and $_2$7-like samples without charge identification, with Gd-doping explicitly noted as being under study in one NSI paper. A representative configuration uses 1 GeV bins in $_2$8 GeV, 20 bins in $_2$9, 30% energy resolution in the sub-GeV region, 10% at multi-GeV, and 10$\le 10^{-14}$0 zenith-angle resolution, together with pulls for 20% flux normalization, 10% cross-section normalization, 5% detector efficiency, and 5% tilt terms (ESSnuSB et al., 2024, ESSnuSB et al., 25 Aug 2025).

In the standard three-flavor atmospheric program, the ESSnuSB atmospheric study reports that the wrong mass ordering can be excluded at $\le 10^{-14}$1 after about 4 years and at $\le 10^{-14}$2 after about 10 years regardless of the true ordering. The wrong $\le 10^{-14}$3 octant can be excluded at $\le 10^{-14}$4 after about 4 years for normal ordering and after about 8 years for inverted ordering; after 10 years, the 3$\le 10^{-14}$5 allowed ranges are $\le 10^{-14}$6 and $\le 10^{-14}$7 for true normal ordering (ESSnuSB et al., 2024). A related NSI study states that without NSI, atmospheric data at ESSnuSB can resolve mass ordering at $\le 10^{-14}$8 and the $\le 10^{-14}$9 octant at $\nu_\mu \rightarrow \nu_e$00, and that these sensitivities remain robust when one NSI parameter at a time is profiled (ESSnuSB et al., 25 Aug 2025).

The NSI analysis formulates propagation in matter with the flavor-basis Hamiltonian

$\nu_\mu \rightarrow \nu_e$01

where only differences of diagonal terms are observable (ESSnuSB et al., 25 Aug 2025). For a 5.4 Mt$\nu_\mu \rightarrow \nu_e$02year exposure and normal ordering, with phases profiled, the resulting 90% CL bounds are $\nu_\mu \rightarrow \nu_e$03, $\nu_\mu \rightarrow \nu_e$04, $\nu_\mu \rightarrow \nu_e$05, $\nu_\mu \rightarrow \nu_e$06, and $\nu_\mu \rightarrow \nu_e$07 (ESSnuSB et al., 25 Aug 2025).

The atmospheric program therefore serves a distinct function within ESSnuSB. It is not merely ancillary to the beam program: it provides independent leverage on matter effects, ordering, octant, and NSI, and the literature repeatedly characterizes it as complementary to the accelerator measurements.

6. Solar and supernova neutrinos, systematic control, and unresolved details

The beam-program proceedings explicitly identify three non-beam channels accessible to the far detector: solar neutrinos with $\nu_\mu \rightarrow \nu_e$08 MeV via elastic $\nu_\mu \rightarrow \nu_e$09–$\nu_\mu \rightarrow \nu_e$10 scattering, atmospheric neutrinos with $\nu_\mu \rightarrow \nu_e$11–10 GeV through charged-current and neutral-current interactions, and supernova-burst neutrinos with $\nu_\mu \rightarrow \nu_e$12–30 MeV through inverse beta decay and $\nu_\mu \rightarrow \nu_e$13–O interactions (Ghosh, 16 Jun 2026). For these channels, the beam proceedings do not quote explicit thresholds beyond those ranges, nor event-rate estimates (Ghosh, 16 Jun 2026).

A dedicated supernova study develops the core-collapse program in much more detail for a 538 kt water-Cherenkov far detector. There, the primary detection channel is inverse beta decay,

$\nu_\mu \rightarrow \nu_e$14

which accounts for about 90% of supernova events in the detector model. The study assumes a benchmark signal efficiency of 95%, a Gaussian energy resolution $\nu_\mu \rightarrow \nu_e$15, and an energy window of 0.5–100 MeV binned into 200 true-energy and 200 reconstructed-energy bins; backgrounds are neglected for a Galactic supernova burst in underground operation (ESSnuSB et al., 25 Jun 2026). The event-rate calculation is written as

$\nu_\mu \rightarrow \nu_e$16

with the fluence diluted by the geometric factor $\nu_\mu \rightarrow \nu_e$17 (ESSnuSB et al., 25 Jun 2026).

At $\nu_\mu \rightarrow \nu_e$18 kpc, the study reports total IBD event counts that depend strongly on the assumed flux model and on the neutrino mass hierarchy. For the Garching model the totals are 51,742 events in normal hierarchy and 53,120 in inverted hierarchy; for GVKM, 87,548 and 85,559; for Livermore, 163,462 and 193,473 (ESSnuSB et al., 25 Jun 2026). With 5% normalization and energy-scale systematics, the same analysis finds strong model discrimination up to distances $\nu_\mu \rightarrow \nu_e$19–$\nu_\mu \rightarrow \nu_e$20 kpc at $\nu_\mu \rightarrow \nu_e$21, while varying the efficiency from 60% to 100% changes $\nu_\mu \rightarrow \nu_e$22 approximately in proportion to efficiency and varying $\nu_\mu \rightarrow \nu_e$23 from 5% to 25% changes $\nu_\mu \rightarrow \nu_e$24 by less than 10% (ESSnuSB et al., 25 Jun 2026).

Systematic control is a recurring theme in all ESSnuSB far-detector studies. Conservative beam analyses often assume 5% signal normalization uncertainty and 10% background normalization uncertainty, sometimes with 5% shape distortions per energy bin and 2% energy-scale uncertainty (Alekou et al., 2021, Fanourakis et al., 21 Jan 2025). ESSnuSBplus proceedings introduce a more aggressive systematic framework in which absolute flux uncertainty is targeted at 1% through monitored-beam instrumentation and a 250 m near detector, cross-section uncertainties are constrained to 3–5% using near detectors and external data, detector model and energy scale are calibrated to 1%, and background modeling is controlled at the 1–2% level using sidebands (ESSnuSB et al., 15 Jan 2026).

Two recurrent misconceptions are corrected by the literature itself. First, the far detector is not described by a single immutable geometry: depending on the study, it is treated as one 538 kt tank, two 270 kt modules, or four 135 kt modules. Second, not every detector parameter is fixed in all publications: some proceedings explicitly do not provide numerical energy-resolution curves, angular-resolution values, detailed PMT coverage, or yearly appearance/disappearance event rates (Ghosh, 16 Jun 2026). For that reason, the far detector is best understood as a stable experimental concept—a megaton-class underground water-Cherenkov system at 360 km optimized for second-maximum oscillation physics—with several engineering and simulation realizations currently coexisting in the arXiv record.