In the depths of the Mediterranean Sea, far from the bright clear blue sky, lies a hidden treasure. It’s not a shipwreck or a pirate’s hoard, a lost artefact or a water-carved sculpture. In fact, at first glance it simply looks like some very organized and oddly stationary bubbles. But these aren’t just trapped pockets of air. They are glass spheres connected by lines of cables, rooted to the sea floor. Swaying slightly with the currents, this odd array is completely alien to the beautiful ocean environment.

Rather than being an art installation seen only by sea creatures and submarines, the unexpected sculpture is a neutrino detector, known as ANTARES (Astronomy with a Neutrino Telescope and Abyss environment RESearch). Tiny and chargeless, neutrinos can be produced artificially in nuclear reactors or created when cosmic rays (protons or heavier nuclei) hit the atmosphere. But physicists at ANTARES are more interested in neutrinos from much further afield, such as remote galaxies. In that case, they are produced when cosmic rays get accelerated and collide with the dense ambient medium.

Unlike charged particles, neutrinos are not deflected by the magnetic fields that permeate the universe; in addition, they interact so weakly with matter that they can travel huge distances across space without being absorbed or scattered. Detecting such neut­rinos and retracing their paths therefore allows the cosmic sources to be pinpointed. These subatomic particles are, however, even harder to detect than those made on Earth because they are so few in number.

The only identified sources of cosmic neutrinos are the Sun and the supernova SN1987A Both were confirmed in the 1980s using, among others, Japan’s Kamiokande detector, which contained 3000 tonnes of ultrapure water in a lab 1000 m below ground. But astrophysicists anticipate much more from the neut­rino sky, especially at energies above 1012 eV (TeV) , which is why they have turned to the oceans. Unlike detectors such as Kamiokande, or its even bigger successor Super-Kamiokande, using the ocean means there’s no need to dig vast underground complexes and no limit to how big the detectors can be.

One beauty of the ocean’s waters is that they serve as a natural shield against the background charged particles (mainly muons) created from cosmic rays interacting in the atmosphere. To further reduce this contamination, neutrino telescopes also concentrate their observations on upward-going neutrinos that have passed through the Earth. These telescopes, in other words, observe the sky on the other side of the Earth, using the planet as a giant particle “filter” that lets only neutrinos through.

But the ocean’s main appeal for physicists is the water itself, which transforms the sea into a giant telescope. In particular, it detects the “Cherenkov light” produced by charged particles that are created when a neutrino interacts with an atom’s nucleus. Moving faster than the speed of light in water, these particles create a cone of blue light at a well-defined angle with respect to the particle’s direction of travel (figure 1) – a process similar to the creation of sound shock waves. In a dark, transparent environment this Cherenkov light can be detected by photomultipliers and then used to reconstruct the energy and incoming direction of the parent neutrino. While both Kamiokande and Super-Kamiokande rely upon this principle, their water tanks are not big enough to detect the extremely faint flux of cosmic neutrinos. The ocean, however, does not have that limitation.

Looking through the Earth

The far-fetched idea of sticking a neutrino detector at the bottom of the sea was first proposed in 1960 by Soviet physicist Moisey Markov, but it was not until the 1970s that the US began work on the first submarine neutrino telescope off the coast of Hawaii – the Deep Underwater Muon and Neutrino Detector (DUMAND). As it was in the northern hemisphere, the detector was designed to find neutrinos from the southern sky, on the opposite side of the planet. In that direction is the inner region of our galaxy, which is known to host a supermassive black hole and plethora of other particle acceleration sites that could be producing cosmic neutrinos.

DUMAND’s planned set-up – and deep-sea detectors installed since – included an array of vertical cables, several hundred metres in height. Also known as “lines”, they were to be anchored into the sea floor at a depth of 4800 m and held in place vertically by immersed buoys. These lines would support clusters of photomultipliers protected from the ocean pressure in centimetre-thick glass spheres about half a metre in diameter. This array was to be connected to the coast through a long electro-optical cable, powering the detector and providing optical-fibre support for data transmission.

But the technological challenges were tremendous and DUMAND was never completed. Working underwater means dealing with high pressure, corrosion and leaky connectors – and you can’t just dive down to do repairs. From 1982 until 1987, some 14 R&D operations at sea were required before the first autonomous-prototype line managed to detect atmospheric muon trajectories, validating the Cherenkov-based detection principle and triggering the installation of the main cable. In December 1993 the first line was connected, but a pressure vessel leak occurred a few hours later, eventually generating a short circuit and causing communication with the installed apparatus to be lost.

That same year also saw scientists begin installing a telescope in Lake Baikal, Siberia – the world’s deepest lake and largest body of fresh water, reaching depths of about 1600 m. The lake gets covered in winter with a thick layer of ice, which made it easier to install the detector because it could carry the weight of heavy instruments without cracking, and an eight-line detector with 192 photomultipliers was quickly deployed. However, the lake water, despite being among the purest in the world, was not ideal for detecting neutrinos because it absorbs light more than ice or sea water. The detection lines therefore had to be placed relatively close to each other, restricting the detection volume to a modest ~5 Mt (equivalent to 0.005 km3).

Back in the US, the DUMAND project was stopped in 1995 due to a lack of funding, and activities were redirected to the installation of a similar detector in the Antarctic ice: the Antarctic Muon And Neutrino Detector Array (AMANDA). As with Lake Baikal in winter, the solid ice made life easier by allowing researchers to drill holes into the ice using hot water, without any need for a ship. The simplified construction partly compensated for performance losses caused by the less favourable optics of ice compared to seawater and the fact the detector was in the southern hemisphere so did not have the Milky Way’s centre in its field of view.

AMANDA stopped operating in 2004 and was upgraded to the famous IceCube Neutrino Observatory. In 2013 this 1 km3-sized detector identified cosmic neutrinos from the depths of space, for which it won Physics World’s Breakthrough of the Year Award in 2013. But the origin of IceCube’s cosmic signal remains unknown. Researchers have been unable to identify the sources because they have limited statistics and directional reconstruction power. The latter limitation is partly due to the important diffusion of light in ice, which degrades IceCube's angular resolution – a crucial parameter for astronomy.

The lure of the Mediterranean

So we return to warmer climates and ANTARES. After the failure of DUMAND, Europe had taken up the torch of submarine neutrino telescopes, concentrating on the Mediterranean Sea because it lies in the northern hemisphere and offers deep-sea sites relatively close to on-shore facilities. After several site surveys and prototype developments in Greece, Italy and France, the ANTARES project began in 1996 off the coast of Toulon, France.

But success was not immediate, with dozens of autonomous tests having to be carried out to understand the environmental conditions. The researchers needed to know everything from how marine currents bend the detection lines and how salinity affects the speed of sound in water (vital for calibration), to how bioluminescence would affect the photomultipliers, and whether biofouling could potentially soil the detection modules. Some brave physicists even occasionally had the privilege of diving to depths of almost 2500 m on board the Nautile – a manned submarine of the French research institute Ifremer that also served for the exploration of the wreck of the Titanic. Among their missions, these researchers had to electro-optically connect the main cable and perform visual inspections of the area. Their adventures even led to some surprises, such as the discovery of an old cannon found lying near the detector.

The first detection line was installed in 2006 and ANTARES was completed in 2008 with 12 detection lines, each featuring 75 photomultiplier spheres along their 450 m and anchored at a depth of 2500 m. ANTARES has so far observed more than 10,000 neutrino events with energies ranging from 100 GeV to several hundreds of TeV. These detected events are compatible with the predicted neutrinos created by the interaction of cosmic rays in the atmosphere, but presumably hide a handful of cosmic neutrinos.

ANTARES, with its unmatched pointing power and good coverage of the central region of our galaxy, is providing important results and complementary information to IceCube. It is also part of an ambitious “multimessenger” programme that seeks to correlate the neutrino events with other cosmic probes, including photons (from radio to gamma rays) and even the recently detected gravitational waves. Despite these efforts, and the presence of a slight excess of events in ANTARES data that could correspond to a cosmic signal, no attempts have succeeded in identifying a neutrino source so far. Confirmation can only come from an even larger detector: the Cubic Kilometre Neutrino Telescope (KM3NeT), which will be the successor of ANTARES.

The next generation: KM3NeT

Construction of KM3NeT began in 2015, with 240 scientists in 15 different countries embarking on the latest deep-sea adventure. And it will be massive. On completion in the early 2020s it will have 345 detection lines distributed across two sites in the Mediterranean Sea: one near Toulon, close by ANTARES, and a second one off the coast of Capo Passero in Sicily, Italy, creating a telescope with a detection volume of more than 1 km3. Subject to future funding, there may also be a third site off the coast of Pylos, Greece.

While neutrino detection in KM3NeT will still be reliant on the Cherenkov principle, the new project features significant technological improvements based on the decade-long experience of ANTARES and the other prototypes. In particular, 31 small photomultipliers – instead of a single, larger one – will be housed in each glass sphere offering several advantages in terms of photon-detection efficiency, photon counting and directionality – all of which are crucial ingredients for the reconstruction of the incoming neutrino energy and arrival direction.

The deployment procedure has also been redesigned: the full detection line is coiled into a spherical frame and attached to a line anchor, which in turn is equipped with an acoustic receiver. Researchers can acoustically monitor the descent of the detection unit from a surface vessel, allowing lines to be positioned to within 1 m. And there’s no need any more for courageous divers as the anchor is connected to the seabed network by a submarine vehicle remotely operated from the boat. Once the connection is verified onshore, an acoustic signal triggers the unfurling of the unit. The compact line frames also mean several lines can be deployed during a single cruise, saving time and money.

Although both detection sites will be based on the same technology, the two will pursue different physics goals. In Toulon, the emphasis will be on studying atmospheric neutrino properties in the GeV energy range, with a dense detector named Oscillation Research with Cosmics in the Abyss (ORCA). In Sicily, a larger and sparser detector called Astrophysics Research with Cosmics in the Abyss (ARCA) will focus on the study of astrophysical sources with energies ranging from TeV to PeV. At each site, the first lines of the arrays have been installed and the first background events have been observed.

Physicists working on the construction of KM3NeT are eager to share their data and provide new opportunities for earth and sea sciences through their cabled infrastructure. From oceanography to geophysics and from marine biology to climatology, the full scientific potential of deep-sea neutrino observatories is still to be explored. As a partner of the European Multidisciplinary Seafloor and water column Observatory facility, KM3NeT will help scientists understand the complex interaction between the geosphere, the biosphere and the hydrosphere, while continuing to hunt for cosmic neutrinos (see box below). So come 2020, there will be many more orderly glass bubbles confusing the fish while probing the depths to look afar.

Box: A multidisciplinary observatory of the sea

It is essential to calibrate and monitor the response of any undersea neutrino detector by measuring environmental parameters such as the optical properties of water, sea currents, bioluminescence and acoustic noise. This is why deep-sea neutrino telescopes not only scrutinize the cosmos from the abyss, but also contribute to more down-to-Earth research. Thanks to their permanent connection with an on-shore lab, facilities such as ANTARES and KM3NeT are providing new and valuable data for oceanographers, geophysicists and biologists, who usually rely on autonomous stations with limited data storage capacity. The long-term and real-time monitoring of deep-sea parameters, such as temperature, pressure, salinity and oxygen and carbon-dioxide content, will allow a better understanding of the marine environment and ecosystems, and of the impact of climate change on the oceans. The deployment of a seismometer and pressure gauges on the site will also contribute to the monitoring and early warning of seismic hazards such as earthquakes and tsunamis. Even the neutrinos registered by KM3NeT, having traversed the whole Earth, can be exploited by geophysicists to obtain indirect information on the composition of the innermost regions of the planet, complementing insights inferred from seismic waves.

The telescopes’ photomultipliers are also sensitive to the continuous background of bioluminescent light emitted by micro-organisms. Although physicists view this light as noise, to marine biologists it’s a valuable signal. One study, for example, combined data obtained by ANTARES and two independent mooring lines also located in the north-west Mediterranean Sea. Looking for correlations between temperature, current velocity and optical activity, the study has led to a better understanding of the link between bioluminescence and the mechanisms of deep-water formation: cold winters densify the surface waters that sink into the abyss by gravity, bringing oxygen and triggering a firework of bioluminescence around ANTARES.

Other discoveries have come from prototype hydrophone arrays in both ANTARES and the Italian neutrino telescope prototype NEMO. These arrays are primarily meant to study the possibility of enhancing the detection of highly energetic neutrinos by listening to their associated sound wave. But they also turn out to be a non-invasive way of monitoring the presence and activities of dolphins and other sea mammals by detecting their acoustic emissions, which can propagate for tens of kilometres in seawater. These range from ultrasonar echolocation “clicks” to frequency-modulated whistles used for social communication. Tracking these signals has even ended up revealing the continuous presence at great depths in the Mediterranean of a population of sperm whales much larger than previously inferred from sound recordings conducted close to the surface. Such studies are invaluable for marine biologists to study how dolphins move, feed, capture prey, communicate and mate.