Neutrinos from our backyard: IceCube sees the Milky Way in neutrinos

Paper Title: Observation of high-energy neutrinos from the Galactic plane

Authors: The IceCube Collaboration

Status: published in Science

The IceCube Neutrino Observatory has been searching for the origins of extraterrestrial neutrinos for over 10 years. In 2013, IceCube announced the first detection of astrophysical neutrinos, and has since performed many searches for their origin. In 2017, a high-energy neutrino was detected, which prompted a huge follow-up effort from telescopes around the world. This particular neutrino was found to be coincident with a flaring blazar, TXS 0506+056. Just last year in 2022, a second source of astrophysical neutrinos from NGC 1068 was reported by IceCube. 

All of these neutrino sources originated from outside our Milky Way galaxy—until now. In today’s paper, the IceCube Collaboration reports the first observation of high-energy neutrinos coming from our own galaxy!Scientists have long expected the galaxy to be a source of neutrinos, both because of the gamma-rays we have observed and and how much closer we are to sources inside our own galaxy. Previously, IceCube has been able to set limits on how many neutrinos might be coming from our galaxy, but now, with new methods and more data, astronomers have finally seen evidence of these neutrinos with IceCube.

What is IceCube, and how does it work?

IceCube is a huge detector embedded deep in the Antarctic ice at the South Pole. It detects the light that is produced when neutrinos interact with atoms in the ice, known as Cherenkov radiation. Each time IceCube sees a signature of a particle passing through the detector, it is called an “event.”

Fig 1: An illustration showing the IceCube detector inside the Antarctic ice (shown for comparison is height of the Eiffel Tower). There are 86 total strings suspended in the ice, with a total of 5160 sensors (called Digital Optical Modules or DOMs) on them. The DOMs are deployed between 1450m and 2450m in the ice. Credit: IceCube Collaboration
two events from IceCube. the one on the left is mostly a straight line through the detector, while the one on the right has a large sphere at the center with smaller spheres around it
Fig 2: Two astrophysical IceCube events. Each sphere is one DOM, and its size shows the amount of light seen by that sensor. The colorbar shows the timing information, with red being the first light seen, to green/blue being the latest. Credit: from IceCube Collaboration 2013 (left is event 5 and right is event 9 in that paper.)

Figure 2 shows two types of events that IceCube detects. The first type, referred to as “tracks,” occurs when a neutrino interacts with the ice and produces a muon, which leaves a clear path as it travels through the detector. This path makes it relatively easy to trace the direction the neutrino came from. However, muons from cosmic rays interacting in our atmosphere produce identical signatures in the detector, making them hard to distinguish from their astrophysical counterparts. 

The second type of event, known as a “cascade,” is more spherical in shape, making them harder to trace back to a source. Today’s paper only investigates cascade-type events, because it’s easier to distinguish cascades from the atmospheric background, and thus boosts the signal.

Using machine learning on IceCube events

IceCube sees a lot of events (about 27,000 per second!), and a majority of them are from the atmosphere. About 100 million atmospheric muons and atmospheric neutrinos interact for every 1 astrophysical neutrino that IceCube sees—it’s like searching for a needle in a stack of needles! To sift through all this noise, the authors use tools similar to Convolutional Neural Networks (like the one described in this Astrobite) to select cascade-type events and cut down on background. Then, a Neural Network (NN) is used to infer the properties of selected events. 

Altogether, using these machine learning methods means that there are more astrophysical neutrinos in the sample, and also improves the certainty on which direction the event came from. In today’s paper, the researchers apply these methods to search for neutrino sources, and using this new dataset they are finally able to detect the neutrinos coming from our galaxy!  

Galactic Neutrinos

The authors did two types of searches: one looking at diffuse emission from the galaxy as a whole, and one looking at individual sources in our galaxy. The first search uses what is called a “template”: a model expectation for the expected flux of neutrinos over the entire Galactic plane. 

In the template search, they use three templates derived from gamma-ray observations. When particles are accelerated by high energy sources, they can interact to produce pions. Neutral pions decay to produce gamma-rays, while charged pions produce neutrinos (and other particles) when they decay. We expect both types to be produced together, so we can use the gamma-rays we see to predict where the neutrinos are coming from. Using the models, the authors find neutrinos coming from the entire galaxy, at high statistical significance (4.48 sigma). 

5 panels showing the galaxy in photons, neutrino expectations, and the neutrinos observed.
Fig 3: The Galactic plane in photons and neutrinos. The top two panels show the Galactic plane as astronomers have observed it in optical and gamma-ray photons. The third panel is the predicted neutrino emission from the Galactic plane. The fourth panel shows the expectation for what IceCube might see from the expectations shown in the third panel, and the fifth and final panel shows the new observations from IceCube of Galactic neutrinos (Figure 1 in the paper)

The second type of search looks at different possible sources (for example, supernova remnants) and combines the signal from all of the known sources of this type. This technique is called “stacking”, since we combine (stack) the signals from all the sources. 

The researchers find that all of these stacking searches are statistically significant, but this could be because many of these sources are also in regions that we expect to see neutrino emission from the template searches.

Galactic Neutrino Astronomy

This long-anticipated result opens the door for Galactic neutrino astronomy, and provides insight into the nature of our own galaxy. But, there are many open questions left – which model best describes the emission? Are there individual sources producing neutrinos within the galaxy, or is the whole flux from diffuse emission? With exciting upgrades to IceCube in the form of IceCube-Gen2 and the IceCube Upgrade, and more years of data, scientists will be better able to probe the answers to these questions. 

Disclaimer: The author of this Astrobite is a member of the IceCube Collaboration and a co-author on this paper.

Astrobite edited by: Briley Lewis

Featured image credit: IceCube Collaboration

About Jessie Thwaites

Jessie is a PhD student at the Wisconsin IceCube Particle Astrophysics Center at the University of Wisconsin-Madison. They study possible astrophysical sources for high energy neutrinos through multimessenger astrophysics. Outside of physics, they play horn and enjoy spending time outdoors, especially skiing and biking.

Leave a Reply