HALO: A Lead Detector Built for One Job

When a massive star’s iron core collapses, almost all of its gravitational binding energy — some $10^{53}$ erg — comes out in neutrinos over the first ten seconds. Every flavor and antiflavor is produced, with characteristic luminosities and average energies that depend on the deleptonization stage, the accretion phase, and the proto-neutron-star cooling phase. The next galactic supernova will be one of the most informative single events in particle astrophysics, and getting the most out of it requires multiple detectors with complementary flavor sensitivities.

Most of the existing detectors are good at one channel: inverse beta decay on free protons, ${\bar{\nu }}_e+p\to e^{+}+n$, which dominates the response of Super-Kamiokande, IceCube, KamLAND, Borexino, and the future SK-Gd. This channel is excellent for electron antineutrinos but blind to electron neutrinos and to the heavy-lepton flavors. To decompose the burst by flavor — which is essential for testing collective oscillation models, the explosion mechanism, and the proto-neutron-star cooling — different detector technologies have to fill in the missing flavors.

HALO, the Helium And Lead Observatory at SNOLAB, is one of the experiments that fills the gap. It is small by modern neutrino-detector standards — 79 tons of lead, 128 helium-3 proportional counters — but it is built specifically for one job: to record a galactic supernova burst in channels other detectors cannot reach. This post is about why lead, how HALO works, and what it adds to the next supernova measurement.

Why lead?

Lead’s appeal as a neutrino target for supernova energies rests on several specific nuclear properties.

The first is the cross-section. At the 10-30 MeV scale typical of supernova neutrinos, ${}^{208}$Pb provides a substantial cross-section for both charged-current and neutral-current interactions, scaling roughly with the squared charge of the nucleus. Per nucleon, lead is actually no more interactive than oxygen or carbon; but per gram, with 82 protons and 126 neutrons in each nucleus, the integrated cross-section per kilogram of target is high.

The second is the dominant decay channel. After a neutrino-induced charged-current or neutral-current event, the daughter nucleus is left in an excited state. For ${}^{208}$Pb the excitation energies of the relevant states sit above the neutron emission threshold but below the threshold for other decay channels, so neutron emission overwhelmingly dominates the de-excitation. Each interaction therefore liberates one or sometimes two neutrons. The neutrons are easy to detect — much easier than the underlying nuclear gamma-ray cascade — through a downstream thermal-capture step that produces a clean signature.

The third is cost and radiopurity. Lead is one of the cheapest dense materials available at multi-ton scale, and ancient lead from sunken Roman ships or carefully aged smelter stock has remarkably low background levels from ${}^{210}$Pb (a daughter of ${}^{222}$Rn). This combination — cheap, low background, neutron-emitting target — does not occur for any other major candidate material at supernova-neutrino energies.

How HALO works

The HALO detector is a $3.4\times 3.7\times 4.0$ meter array of 79 tons of lead arranged around 128 cylindrical helium-3 filled proportional counters. The counters were originally manufactured for the Sudbury Neutrino Observatory (SNO) as part of its neutral-current detector phase, where they detected free neutrons from $\nu +d\to n+p+\nu$ in heavy water. After SNO completed its third phase in 2006, the counters were repurposed for HALO and embedded in the lead block.

The detection chain has two steps:

1. A supernova neutrino interacts in the lead. Depending on the channel, one or two neutrons are emitted from the excited lead nucleus.
2. Each emitted neutron thermalises through scattering in the lead and surrounding moderator material, then is captured by a ${}^3$He nucleus in one of the proportional counters: $n+{}^3$He$\to {}^3$H + $p$, releasing 764 keV of kinetic energy shared between the triton and the proton. The two charged particles produce ionisation that the proportional counter reads out as a clean signal.

The total event-detection time per interaction is approximately 200 microseconds — fast enough to handle the burst rate from a galactic supernova (estimated at hundreds to thousands of events over the burst’s 10-second duration for the nearest plausible supernovas).

The energy of each detected event is not separately measured beyond the neutron-multiplicity counting. HALO records, per neutrino interaction, whether one or two neutrons were emitted, and the time of each. Combined with the directional and temporal cuts from coincident detectors in the SNEWS network, this is enough to identify the burst event and characterise its flavor content.

One neutron versus two: the spectral handle

The key piece of information HALO records beyond simple event counting is the neutron multiplicity per interaction. The reason this is diagnostic comes from the underlying nuclear structure.

A charged-current event with an electron neutrino, ${\nu }_e{+}^{208}$Pb$\to e^{-}{+}^{208}$Bi${}^{\ast }$, leaves the bismuth nucleus in an excited state whose energy depends on the incident neutrino energy and the particular nuclear transition. Low excitation energies ($\sim 7$ MeV above the bismuth ground state) are above the one-neutron emission threshold but below the two-neutron threshold; higher excitations open up the two-neutron channel as well.

The result is a definite branching ratio between one-neutron and two-neutron events that depends on the neutrino energy spectrum. A spectrum dominated by higher-energy neutrinos produces more two-neutron events; a softer spectrum produces only one-neutron events. By counting the ratio over the burst, HALO measures a coarse-grained spectral feature without needing an energy-resolved detector.

The same logic applies to neutral-current events on ${}^{208}$Pb, with somewhat different multiplicity branching ratios because the neutral-current operator excites different nuclear states.

What it adds at a galactic supernova

For a supernova at the typical distance of 10 kiloparsecs, HALO expects to record roughly 30 to 50 events over the 10-second burst, depending on the specific supernova model. This is far fewer than Super-K (~10,000 events), IceCube (~1 million events through its diffuse signal), or DUNE (~3,000 events through CC on argon), but the events HALO records are in channels the larger detectors cannot fully access.

Super-K’s water-Cherenkov sample is dominated by inverse beta decay on protons — about 7,000 events out of the 10,000 — leaving only a small fraction in other channels. IceCube’s signal is integrated over the whole ice volume and provides almost no event-by-event flavor information. DUNE’s LArTPC dominates the electron-neutrino charged-current channel through ${\nu }_e+{}^{40}$Ar$\to e^{-}+{}^{40}$K${}^{\ast }$, complementary to inverse-beta-decay-driven detectors. HALO’s CC and NC measurements on lead provide a third probe that depends differently on the cross-section uncertainties and on the burst-spectrum modelling.

When all the detectors are combined in a global fit to a single galactic supernova, the flavor content of the burst — the relative luminosities of ${\nu }_e$, ${\bar{\nu }}_e$, and the combined ${\nu }_x$ (where $x$ stands for ${\nu }_{\mu },{\nu }_{\tau }$, and their antiparticles) — can be reconstructed at much higher precision than any single detector achieves on its own.

HALO and SNEWS

Since 2012 HALO has been one of the four primary detectors in the Supernova Early Warning System (SNEWS), the international network that aims to issue a coordinated early warning when a galactic supernova is detected. The other members are Super-Kamiokande, IceCube, and KamLAND. LVD at LNGS, Borexino, Daya Bay, and NOvA are associate participants. SNEWS combines detector triggers in coincidence to produce a high-confidence supernova alert that can be issued to optical astronomers within minutes of the neutrino burst, allowing electromagnetic follow-up of the explosion before the light from the photosphere can reach Earth (which takes hours for the typical galactic supernova).

HALO’s role in SNEWS is to provide an independent confirmation trigger through its unique flavor sensitivity. Even if its event rate is modest, the coincidence of a HALO trigger with triggers from other detectors strengthens the confidence in the alert.

HALO-1kT: the planned upgrade

A roughly twelvefold scale-up is in design as HALO-1kT: a 1-kiloton lead target with a correspondingly larger neutron-detection system. The lead would be located either at SNOLAB or at another deep underground site, with the expected event rate from a galactic supernova at 10 kpc reaching about 500 events. The larger sample would allow finer spectral binning of the one-neutron-versus-two-neutron ratio and substantially improved flavor decomposition.

The technical challenges of HALO-1kT are largely logistical rather than scientific. Procuring 1 kiloton of radiopure aged lead, designing a neutron-detection system that scales to the required volume, and finding a deep underground location with the necessary space and ancillary infrastructure are the main engineering goals. The science case is clear: an order-of-magnitude expansion of HALO’s unique capability, ready in time for the (probable) next galactic supernova.

Why a dedicated detector matters

The next galactic supernova will happen, on average, somewhere between every 30 and 100 years — Galactic SN1987A was 39 years ago, suggesting we might be due. When it does, the burst will release approximately $10^{58}$ neutrinos, of which the various detectors at Earth will record perhaps a few hundred thousand in total. Each detector technology contributes a different piece of the puzzle, and the absence of any one piece would significantly limit what can be reconstructed.

HALO sits at the small-but-essential end of this network. Its events are few compared to the giant Cherenkov and scintillator detectors, but they are information-rich in a way the giant detectors are not. The ratio of one-neutron to two-neutron events directly measures part of the supernova spectrum that no other detector accesses easily. The charged-current sensitivity to ${\nu }_e$ — through CC on ${}^{208}$Pb — complements the inverse-beta-decay sensitivity to ${\bar{\nu }}_e$ that dominates the larger samples. Without HALO (and its scaled successor) the flavor decomposition of the burst would be significantly less constrained.

Summary

HALO is a dedicated supernova-neutrino detector at SNOLAB built around 79 tons of lead and 128 helium-3 proportional counters salvaged from SNO. Neutrinos from a galactic supernova interact with ${}^{208}$Pb through charged-current and neutral-current channels, producing excited bismuth or lead nuclei that de-excite by emitting one or two neutrons. The neutrons are detected through capture on ${}^3$He in the embedded proportional counters. The ratio of one-neutron to two-neutron events constrains both the flavor content and the spectral shape of the burst in ways no Cherenkov or scintillator detector reaches. HALO has been continuously operating in the SNEWS network since 2012, and the planned HALO-1kT expansion would scale the technique by an order of magnitude in time for the next galactic supernova. Small, dedicated, and built for one job, HALO represents a complementary slice of the supernova-neutrino detection landscape that the giant general-purpose detectors cannot replace.