您当前的位置: 首页 > 网页快照
A tidal disruption event coincident with a high-energy neutrino
Abstract . Cosmic neutrinos provide a unique window into the otherwise hidden mechanism of particle acceleration in astrophysical objects. The IceCube Collaboration recently reported the likely association of one high-energy neutrino with a flare from the relativistic jet of an active galaxy pointed towards the Earth. However a combined analysis of many similar active galaxies revealed no excess from the broader population, leaving the vast majority of the cosmic neutrino flux unexplained. Here we present the likely association of a radio-emitting tidal disruption event, AT2019dsg, with a second high-energy neutrino. AT2019dsg was identified as part of our systematic search for optical counterparts to high-energy neutrinos with the Zwicky Transient Facility. The probability of finding any coincident radio-emitting tidal disruption event by chance is 0.5%, while the probability of finding one as bright in bolometric energy flux as AT2019dsg is 0.2%. Our electromagnetic observations can be explained through a multizone model, with radio analysis revealing a central engine, embedded in a UV photosphere, that powers an extended synchrotron-emitting outflow. This provides an ideal site for petaelectronvolt neutrino production. Assuming that the association is genuine, our observations suggest that tidal disruption events with mildly relativistic outflows contribute to the cosmic neutrino flux. You have full access to this article via your institution. Download PDF Download PDF Main . On 2019 October 1, the IceCube Neutrino Observatory 1 reported the detection of a ~0.2?PeV neutrino, IC191001A, with an estimated 59% probability of being of astrophysical origin solely on the basis of reconstructed energy 2 . Seven hours later, the direction of the incoming neutrino was observed by the Zwicky Transient Facility (ZTF) 3 as part of our neutrino follow-up programme. The data were processed by our multimessenger pipeline ( Methods ), which performs searches for extragalactic transients in spatial and temporal coincidence with high-energy neutrinos 4 , and the radio-emitting tidal disruption event (TDE) AT2019dsg was identified as a candidate neutrino source. TDEs are rare transients that occur when stars pass close to supermassive black holes. Studies have suggested that TDEs are sources of high-energy neutrinos and ultra-high-energy cosmic rays 5 , 6 , 7 ; this holds in particular for the subset of TDEs with relativistic particle jets 8 , 9 , 10 , 11 . Those TDEs with non-thermal emission are considered the most likely to be sources of high-energy neutrinos. AT2019dsg was thus quickly identified as a promising candidate neutrino source 12 . Given that there are typically ? 2 radio-emitting TDEs in the entire northern sky at any one time, we find that in the 80?sq.?deg. of sky observed during the eight neutrino follow-up campaigns by ZTF up to March 2020 the probability of finding a radio-detected TDE–neutrino association by chance is <0.5%. With the second-highest bolometric energy flux of all 17 TDEs detected by ZTF, the probability of finding a TDE at least as bright as AT2019dsg by chance is just 0.2%. These calculations are valid for any isotropic distribution, and therefore quantify the probability that the AT2019dsg–IC191001A association would arise from atmospheric backgrounds. Our programme targets four neutrino population hypotheses 13 , of which the greatest sensitivity is for TDEs ( Methods ). Thus, although not directly reflected in the calculation, the impact of multiple hypothesis tests on these estimates would be modest. While an atmospheric origin for the IC191001A–AT2019dsg association cannot be excluded, the improbability of chance temporal and spatial coincidence substantially reinforces the independent energy-based evidence of an astrophysical origin for IC191001A, and indicates that any atmospheric origin is unlikely. AT2019dsg was discovered 14 by ZTF on 2019 April 9, and was classified as a TDE on the basis of its optical spectrum 15 (see Extended Data Fig. 1 ). This spectrum showed a redshift of z ?=?0.051, implying a luminosity distance D L ?≈?230?Mpc assuming a flat cosmology with Ω Λ ?=?0.7 and H 0 ?=?70?km?s ?1 ?Mpc ?1 . The optical/UV continuum of AT2019dsg is well described by a single blackbody photosphere with a near-constant temperature 16 of 10 4.59±0.02 ?K and radius of 10 14.59±0.03 ?cm. The peak luminosity of 10 44.54±0.08 ?erg?s ?1 is in the top 10% of the 40 known optical TDEs to date 16 , and the temperature is in the top 5%. The late-time evolution is consistent with the rapid formation of an accretion disk 17 , 18 (Fig. 1 ), which would be expected on these relatively short timescales for disruptions around higher-mass supermassive black holes. Indeed the total mass of the host galaxy of AT2019dsg is in the top 10% of all optical TDE hosts. Assuming that 50% of the host mass is in the bulge, we estimate 19 a black hole mass of ~3?×?10 7 M ⊙ . Fig. 1: Multiwavelength lightcurve of AT2019dsg. a , The optical photometry in bands g and r from ZTF (in green and red, respectively), alongside UV observations in bands UVW2, UVM2, UVW1 and U from the Neil Gehrels Swift Observatory (Swift)-UVOT (Ultraviolet/Optical Telescope) (in pink, violet, navy and blue, respectively). The left axis shows νF ν , where F ν is the spectral flux density at frequency ν , while the right axis shows νL ν , where L ν is the luminosity at frequency ν . The late-time UV observations show an apparent plateau, which is not captured by a single-power-law decay. The dashed pink line illustrates a canonical t ?5/3 power law, while the dotted pink line illustrates an exponentially decaying lightcurve. Neither model describes the UV data well. b , The integrated X-ray energy flux, from observations with Swift-XRT (X-Ray Telescope) and XMM-Newton, in the energy range 0.3–10?keV. Arrows indicate 3 σ upper limits. The vertical dotted line illustrates the arrival of IC191001A. Error bars represent 1 σ intervals. Full size image AT2019dsg was also detected in X-rays, beginning 37?d after discovery (Fig 1 , see also Extended Data Fig. 2 ). Though the first X-ray observation indicated a bright source, with a high X-ray to optical ratio of L X / L opt ?≈?0.1, this X-ray flux faded extremely rapidly, as shown in Fig. 1 . This rate of decline is unprecedented, with at least a factor of 50 decrease in X-ray flux over a period of 159?d. Similarly to the optical/UV emission, the observed X-ray spectrum is consistent with thermal emission, but from a blackbody of temperature 10 5.9 ?K (0.072?±?0.005?keV) and, assuming emission from a circular disk, a radius of ~2?×?10 11 ?cm (see Extended Data Fig. 3 ). As for most X-ray-detected TDEs 20 , 21 , 22 , the blackbody radius appears to be much smaller than the Schwarzschild radius ( R S ?≈?10 13 ?cm) inferred from the galaxy scaling relation 19 . X-ray emission is generally expected to arise close to the Schwarzschild radius. Small emitting areas can arise from an edge-on orientation, because the relativistic velocities at the inner disk can Doppler boost a large area of the disk out of the X-ray band. Since our observations probe close to the Wien tail of the spectrum, a small temperature decrease due to absorption would also yield a substantially underestimated blackbody radius and luminosity 22 . The exponential decrease of the flux could be caused by cooling of the newly formed TDE accretion disk 18 or increasing X-ray obscuration. Radio observations shown in Fig. 2 reveal a third distinct spectral component, namely synchrotron emission from non-thermal electrons (see also Extended Data Figs. 4 and 5 ). We model this emission with a conical geometry as expected for outflows (for example jets or winds) that are launched from—and collimated by—the inner parts of flared accretion disks that emit close to the Eddington limit. Given that electrons are typically accelerated with much lower efficiency than protons in astrophysical accelerators 23 , we assume that they carry 10% of the energy carried by relativistic protons ( ? e ?=?0.1). We further assume that the magnetic fields carry 0.1% of the total energy ( ? B ?=?10 ?3 ), as indicated by radio observations of other TDEs 24 and supernovae 25 . We note that the opening angle for the outflow is largely unconstrained. For a half-opening angle, ? , of 30° we find R ?=?1.5?×?10 16 ?cm in our first epoch (41?d after discovery), increasing to R ?=?7?×?10 16 ?cm shortly after the neutrino detection (177?d after discovery). These radii scale 26 as R ∝ ?[1???cos( ? )] ?8/19 . The implied expansion velocity is roughly constant at \(v/c=\dot{R}/c=0.12\pm 0.01\) during the first three epochs, with a significant (>3 σ ) acceleration to v / c ?=?0.21?±?0.02 for the last epoch. These are the velocities of the synchrotron-emitting region, and thus provide a lower limit to the velocity at the base of the outflow. Indeed even the hotspots of relativistic jets from active galaxies that are frustrated by gas in their host galaxy are typically observed 27 to have subrelativistic expansion velocities of ~0.1? c . Fig. 2: Synchrotron analysis of AT2019dsg. a , Radio measurements from MeerKAT (1.3?GHz), the Karl G. Jansky Very Large Array (VLA; 2–12?GHz) and the Arcminute Microkelvin Imager (AMI; 15.5?GHz) at four epochs with times listed relative to the first optical detection. The coloured lines show samples from the posterior distribution of synchrotron spectra fitted to the measurements at each epoch, and the dashed lines trace the best-fit parameters for that epoch. The free parameters are the electron power-law index ( p ?=?2.9?±?0.1) and the host baseline flux density, plus the magnetic field and radius for each epoch. b , The energy at each epoch for a conical outflow geometry with an half-opening angle of 30°. The dotted line indicates a linear increase of energy. c , The corresponding radius for each epoch, with a dotted line illustrating a linear increase. Error bars represent 1 σ intervals. Full size image The inferred outflow energy, E , shows a linear increase from 2.5?×?10 49 ?erg to 2?×?10 50 ?erg (Fig. 2 ), which would not be expected from models of TDE radio emission that involve a single injection of energy 28 , 29 . The constant increase of energy implies a constant injection rate at the base of the outflow of approximately 2?×?10 43 ?erg?s ?1 . While some scenarios can yield an increase in inferred energy from a single energy injection, none of these are consistent with the full set of observed properties. First, a single ejection with a range of velocities could explain the observed linear increase of energy with time (the slower ejecta arrive later), but is incompatible with the increasing velocity. Second, an increase of the efficiency for conversion of Poynting luminosity to relativistic particles is unlikely because the target density that is available to establish this conversion is decreasing. Finally, an apparent increase of the inferred energy due to an increase of solid angle that emits to our line of sight is only expected for relativistic outflows that decelerate. Instead, for AT2019dsg, the observations suggest the presence of a central engine that yields continuous energy injection through a coupling of accretion power to the radio emission 30 , with acceleration in the final radio epoch due to a decrease in the slope of the ambient matter density profile. Neutrino emission from AT2019dsg . With this strong evidence for three distinct emission zones derived purely from multiwavelength observations, we consider whether this picture is consistent with AT2019dsg being the source of the neutrino IC191001A. In particular, neutrino production requires protons to be accelerated to sufficiently high energies, and to collide with a suitably abundant target. The detection of a single high-energy neutrino implies a mean expectation in the range 0.05?10?sq.?deg.). This astrophysical probability was not reported for high-energy starting events under the old IceCube alert selection, or for one recent alert, IC200107A, that was identified outside the standard alert criteria 38 . Each neutrino localization region can typically be covered by one or two ZTF observation fields. Multiple observations are scheduled for each field, with both g and r filters, and a separation of at least 15?min between images. These observations typically last for 300?s, with a typical limiting magnitude of 21.0 m . ToO observations are typically conducted on the first two nights following a neutrino alert, before swapping to serendipitous coverage as part of the public survey. Following observations, images are processed by IPAC 39 , and alert packets are generated for significant detections from difference images 40 . This alert stream of significant detections is then filtered by our follow-up pipeline built within the AMPEL framework 41 , a platform for realtime analysis of multimessenger astronomy data 42 . Our selection is based on an algorithm for identifying extragalactic transients 42 . We search ZTF data both preceding and following the arrival of the neutrino. To identify candidate counterparts to the neutrino, we apply the following cuts to ToO and survey data. We reject likely subtraction artefacts using machine learning classification and morphology cuts 43 . We reject moving objects through matches to known nearby solar system objects 39 . We further reject moving objects by requiring multiple detections for each candidate (that is, at the same location) separated temporally by at least 15?min. We remove stellar sources by rejecting detections cross-matched 44 to objects with measured parallax in Gaia Data Release 2 data 45 , defined as non-zero parallax with a significance of at least 3 σ . We further reject likely stars with machine learning classifications 46 , based on sources detected by Pan-STARRS1 47 , removing those objects with an estimated stellar probability greater than 80%. We identify likely active galactic nuclei (AGNs) by cross-matching to the Wide-field Infrared Survey Explorer survey and applying IR colour cuts 48 . We reject detections consistent with low-level AGN variability. We require that objects lie within the reported 90% error region to ensure spatial coincidence, and that they are detected at least once following the neutrino arrival time to ensure temporal coincidence. These cuts typically yield ~0.2 candidates per square degree of sky. Promising candidates are prioritized for spectroscopic classification, to confirm or rule out a possible association with a given neutrino. AT2019dsg (RA[J2000]?=?314.26°, dec.[J2000]?=?+14.20°) was spatially coincident with the 90% localization of the neutrino IC191001A 2 (RA?=?314.08 \({\,}_{-2.26}^{+6.56}\) °, dec.?=?+12.94 \({\,}_{-1.47}^{+1.50}\) °), at a distance of 1.27° from the best-fit position. It was also temporally coincident, being detected by ZTF in our ToO observations following the neutrino detection. There were additionally three candidate supernovae found in the error region of IC191001A, consistent with background expectations. AT2019dsg was the first TDE identified by our pipeline, and the first TDE to be reported in coincidence with any high-energy neutrino. Probability of chance coincidence . During the first 18 months of survey operations, ZTF identified 17 TDEs 16 , distributed over 28,000° of observed sky (the ZTF survey footprint, after removing sources with a Galactic latitude ∣ b ∣ ?