On this page
On 17 August 2017 at 12:41 UTC, the LIGO-Virgo gravitational-wave network recorded a chirp signal lasting approximately 100 seconds. The signal’s frequency increased from a few tens of Hz to several hundred Hz over its duration — the classic signature of a binary inspiral. Analysis of the waveform indicated two compact objects of masses approximately 1.4 solar masses each, consistent with a binary neutron-star (BNS) merger.
Approximately 1.7 seconds later, the Fermi gamma-ray space telescope detected a short gamma-ray burst from the same general sky region. GRB 170817A was unusually weak — about 1000 times less luminous than typical short GRBs — but its timing coincidence with GW170817 made the association compelling.
Within hours, optical telescopes had identified the source: a fading transient in the galaxy NGC 4993 at a distance of 130 million light-years. The optical counterpart was a “kilonova” — the predicted electromagnetic signature of radioactive decay of heavy elements produced by neutron-star merger ejecta.
This was the first time a single astrophysical event had been observed across multiple messengers — gravitational waves, gamma rays, optical light, and (later) radio. The era of practical multi-messenger astronomy had begun. The announcement on 16 October 2017 was, by any standard, a landmark in astrophysics.
But one messenger was missing. Three neutrino observatories — IceCube, ANTARES, and the Pierre Auger Observatory — searched the GW170817 sky region in time windows around the merger. None detected a coincident neutrino. The null result was, in its own way, as informative as the positive detections.
This post is about the GW170817 multi-messenger event and what the absence of a neutrino signal told us.
The merger and its predicted neutrino emission
A binary neutron-star merger is, energetically, one of the most extreme events in the universe. Two compact objects with combined mass of approximately 2.8 solar masses spiral together and merge in roughly a millisecond, releasing approximately 10⁵³ erg of gravitational binding energy. The remnant is either a single neutron star (which may collapse to a black hole within seconds) or a black hole formed promptly.
Several distinct mechanisms could produce neutrinos:
Thermal neutrinos from the merger remnant. The hot, dense matter formed during the merger reaches temperatures of tens of MeV. Like a core-collapse supernova, it produces a burst of low-energy thermal neutrinos (MeV scale) that escape from the merging matter on a timescale of seconds. The integrated energy in thermal neutrinos is expected to be approximately 10⁵² to 10⁵³ erg — comparable to a supernova neutrino burst.
Non-thermal neutrinos from the relativistic jet. The short gamma-ray burst is powered by a relativistic jet, with bulk Lorentz factors of order 100. Protons accelerated in the jet can produce high-energy neutrinos through photoproduction processes (proton-photon interactions producing pions, which decay to neutrinos). The expected neutrino flux depends sensitively on the jet’s energy fraction in protons and on the local photon field.
Disk wind neutrinos. The post-merger accretion disk drives a wind of ejected material that can produce both thermal and non-thermal neutrinos. The contribution is harder to predict but generally smaller than the thermal-merger or jet components.
Predictions for the integrated high-energy neutrino flux from a single BNS merger at 130 million light-years ranged over approximately 5 orders of magnitude depending on the model. Optimistic models predicted detectable signals at IceCube; pessimistic models predicted nothing.
The neutrino searches
IceCube searched for high-energy neutrinos in the GW170817 sky region. Two time windows were used:
- A short window of ±500 seconds around the merger, optimized for catching prompt jet neutrinos
- An extended window of 14 days, allowing for delayed emission from the post-merger evolution
The IceCube analysis (Albert et al., Astrophysical Journal Letters 850, L35, 2017) reported no significant excess of neutrino events in either time window. The upper limits on the per-flavor neutrino fluence at 100 TeV were approximately:
- ±500 s window: erg/cm²
- 14-day window: erg/cm²
ANTARES in the Mediterranean searched for both high-energy and low-energy neutrinos. The low-energy search (at MeV scale) was relevant for the thermal merger neutrinos. ANTARES is below the Northern hemisphere sky where GW170817 was located, giving it good geometric acceptance.
ANTARES reported no significant detection in either energy range. Combined with IceCube, the upper limits constrained the high-energy fluence to approximately erg/cm² at 100 TeV.
Pierre Auger Observatory searched for ultra-high-energy neutrinos (above eV) from the GW170817 region. The Auger acceptance covers a different energy regime than IceCube and ANTARES, so its constraints are complementary at the highest energies.
Auger reported no detection. The combined three-observatory upper limits provided the first multi-energy constraint on neutrino emission from a confirmed BNS merger.
What the null result told us
The absence of a detected neutrino signal does not mean the merger produced no neutrinos. It means the neutrino fluence reaching Earth was below the experimental thresholds. Several interpretations:
The thermal neutrino burst was at MeV energies and below the IceCube/ANTARES threshold. The expected thermal neutrino spectrum peaks around 10-20 MeV, much lower than IceCube’s PeV-scale sensitivity. Even if the total thermal energy was erg, the per-event energy was too low to detect at 130 million light-years.
The high-energy jet was off-axis from Earth. The relativistic jet that produced GRB 170817A was much weaker than typical short GRBs, suggesting that Earth’s line of sight was significantly off the jet axis. For off-axis observers, the high-energy neutrino flux would be similarly suppressed. The radio afterglow observed over the following months confirmed this picture: the jet had structure with a narrow core and a wider, slower wind.
Models predicting detectable high-energy neutrinos required either: (a) higher proton acceleration efficiency in the jet than commonly assumed; or (b) on-axis viewing geometry; or (c) different jet parameters than inferred from electromagnetic observations. The null result constrained these models.
The null result therefore did not require new physics or revision of merger models — it required pointing-geometry effects and constrained the jet efficiency. Combined with the electromagnetic observations, the picture was self-consistent: a relatively weak, off-axis jet plus a substantial isotropic kilonova, but no detectable high-energy neutrino flux.
What it would have looked like
For a thought experiment: had the GW170817 jet been on-axis (Earth in the jet cone), the expected high-energy neutrino flux could have been substantially larger. IceCube’s sensitivity at 100 TeV for a 100-second window suggests detection probability of order few percent to tens of percent for an on-axis GRB at 130 million light-years, depending on jet parameters.
For thermal merger neutrinos at MeV energies, Super-Kamiokande and large liquid-scintillator detectors (KamLAND, JUNO upcoming) are the relevant probes. Super-K reported approximately 1 candidate event in the GW170817 time window, consistent with background. A future BNS merger at substantially closer distance — say, 20 Mpc instead of 130 Mpc — would produce approximately 40 times higher thermal neutrino flux and might be marginally detectable in Super-K-class detectors.
JUNO (commissioning in 2026) will have approximately 100,000-event sensitivity to a galactic supernova at 10 kpc; for a BNS merger at 100 Mpc, the expected event count is approximately 1, consistent with the original GW170817 (no detection). Closer mergers (10-20 Mpc) might give a handful of JUNO events.
The realistic forecast: most BNS mergers in the next decade will not produce detectable neutrino signals. A nearby merger, or a more favorable jet geometry, could change this.
The broader multi-messenger context
GW170817 was a special event because it was the first BNS merger observed across multiple channels. The combined dataset enabled measurements that no single channel could provide:
Hubble constant. The standard-siren measurement (gravitational-wave distance combined with the electromagnetic-counterpart redshift) gives km/s/Mpc, independent of the cosmic-distance-ladder methods.
r-process nucleosynthesis. The optical kilonova spectrum showed lines consistent with r-process products including lanthanides — confirming that BNS mergers contribute to heavy-element synthesis.
Equation of state of neutron-star matter. The gravitational-wave signal at high frequencies constrains the radii of the merging neutron stars, which probe the dense-matter equation of state.
Jet structure of short GRBs. The combined electromagnetic observations constrained the jet structure, opening angle, and viewing geometry.
The neutrino non-detection added one more constraint: the high-energy jet efficiency was bounded by the IceCube limits. Each independent channel contributed to the integrated picture.
What’s next
The Advanced LIGO-Virgo network has been observing additional BNS mergers since 2017. Several candidate events have been reported, but none have produced electromagnetic counterparts as bright and well-localized as GW170817. As detector sensitivity improves through the late 2020s, the BNS detection rate is expected to grow to several per year.
For neutrino observations:
- IceCube-Gen2 (the planned expansion) will have ~8× the volume of current IceCube. For high-energy neutrinos from BNS mergers, this corresponds to ~8× more sensitivity per merger, plus higher event rates over a multi-year programme.
- KM3NeT in the Mediterranean is approaching full deployment. As a Northern-hemisphere observatory, it complements IceCube’s Southern-sky coverage.
- JUNO and Hyper-Kamiokande will provide much-improved low-energy (MeV-scale) sensitivity for thermal merger neutrinos. A relatively close BNS merger (say, within 20 Mpc) could give a few-event neutrino signal in these detectors.
By 2030, the field expects to have observed several BNS mergers electromagnetically. A few of these may produce detectable neutrino signals if the geometry is favorable. The combined dataset will allow statistical constraints on the merger-driven neutrino emission models, even without a single high-significance detection.
A new normal
Before GW170817, multi-messenger astronomy was an aspiration. After GW170817, it is a routine practice. The event demonstrated that gravitational-wave detections can be followed up by electromagnetic and (in principle) neutrino observations to produce integrated pictures of single astrophysical events.
The fact that no neutrino was detected from GW170817 does not diminish the significance of the multi-messenger event. The neutrino null result is itself a data point in the integrated picture. It tells us specific things about the jet geometry and efficiency that the electromagnetic observations alone would not have constrained.
For the broader neutrino-astronomy programme, GW170817 set the precedent. Every future BNS merger will be searched in the IceCube data for coincident neutrinos. Most of these searches will return null results. Some, eventually, will return a positive detection — the first BNS-merger neutrino, on top of the GW + EM observations. That event will be a landmark in its own right.
GW170817 was, in the meantime, the first test run. The instruments worked. The protocols worked. The community responded. The multi-messenger era is here, and neutrinos are part of it — even when the count is zero.
For more on the broader neutrino-astronomy programme, see IceCube and neutrino astronomy, TXS 0506+056 — the blazar neutrino, SNEWS supernova early warning, and KM3NeT Mediterranean neutrino telescope.