The race to detect dark matter 25 Apr 92

From New Scientist magazine, vol 134 issue 1818, 25/04/1992, page 37

Dark matter makes up most of the Universe, but no one knows what it is. Now physicists around the world are competing to detect new exotic particles that could explain all

Galactic dark matter
Galactic dark matter

Three ways of detecting collisions
Three ways of detecting collisions

Astronomy in the physics lab
Astronomy in the physics lab

One of the most puzzling aspects of our Universe is that most of its mass seems to be invisible. The galaxies of stars that we see account for between one and 10 per cent of it. During the past 15 years, it has become clear that a new form of matter called 'dark matter' is needed to explain observations. This has brought together astronomers and particle physicists to try to discover what new objects or subatomic particles could be responsible for it. There is now a worldwide race, in which Britain is a key participant, to detect the dark matter in our Galaxy.

How do we know that dark matter exists? Our own Galaxy provides clear and dramatic evidence. The mass of an object can be calculated from its gravitational pull on another body. For example, by knowing the Moon's speed and distance from the Earth you can work out the gravitational pull needed to keep it in orbit. From this you can calculate the weight of the Earth (6 x 10 21 tonnes). In the same way you can work out the weight of the Sun from the Earth's orbit. But what happens if we try to use the same method to find the weight of the Galaxy?

The Galaxy consists of 100 billion stars - including our Sun - held in a cluster by gravity. It rotates about its centre, so the outer stars and gas must be held in place by the gravitational pull of the rest. From this you can calculate the total mass in the Galaxy. But this time there is something seriously wrong. The gravitational force needed is 10 times as large as that which could result from the stars alone.

Similar studies of other galaxies produce the same puzzling results. Moreover, the galaxies attract each other into clusters, and from their motion we can calculate that there is still more unidentified dark matter between the galaxies. So astronomers are forced to conclude that the Universe must contain a huge amount of invisible matter. At the moment there is no way of telling what this dark matter consists of. Some astronomers think it could be in the form of large objects such as dead stars or black holes. Others have suggested that the dark matter could consist of vast clouds of subatomic particles. Here, particle physicists have offered several suitable candidates.

The first possibility considered, in the early 1980s, was the neutrino. These light, neutral particles are known to exist and would have been created in large numbers at the beginning of the Universe. We still do not know if the neutrino has a mass, but even if one type of neutrino has a mass about one ten-thousandth of an electron this would be sufficient (because there are, on average, about 100 neutrinos per cubic centimetre throughout all space) to account for all the dark matter in the Universe. However, calculations then seemed to show that it was difficult to explain how galaxies would form in the presence of large numbers of neutrinos.

In 1983, particle theorists suggested two other dark matter candidates. One was a new neutral particle called the axion, less than one-millionth of the mass of an electron. The other possibility was a stable neutral particle that was heavier than the neutron but which interacted only feebly with ordinary matter. The second idea is regarded by many theorists as more likely, and is also more popular with the experimentalists who would like to look for it. This type of new particle is known as a WIMP (weakly interacting massive particle). Cosmologists have calculated that enough WIMPs could have been formed in the early Universe to account for the high proportion of dark matter. Furthermore, WIMPs would have been moving slowly enough for gravity to have gathered them into a cloud or 'halo' of dark matter around each galaxy.

Winkling out WIMPs

Theorists have suggested other ideas - for example, the simple law of gravity could be wrong when applied to something as big as a galaxy. Most cosmologists think this is very unlikely (New Scientist, Science, 8 February) but nevertheless it cannot yet be totally ruled out.

By 1983, several research groups around the world, including particle physicists at the Rutherford Appleton Laboratory, had begun to think about experiments that might detect the dark matter. Some experiments - particularly those for detecting neutrinos - were not technically possible in the immediate future, but by 1985 it became clear that experiments to detect WIMPs would certainly be feasible.

How would we detect WIMPs? If they fill the Galaxy, they must be all around us. From the calculated density of dark matter in the Galaxy, astronomers estimate that there would be only about 10 or 100 in every litre of space, but that they would be moving at one-thousandth of the speed of light, so there would be about a million of them streaming through every square centimetre of area each second. This means that around 1015 WIMPs would pass through a human body each day. Yet these particles are so feeble that only around 100 a day would interact with atoms in your body. The vast majority would pass through us and the Earth unaffected.

However, just a few will collide with atoms. About one WIMP a day would collide with an atom in a kilogram of any material, and that is the key to detecting them. If a moving billiard ball collides with a stationary ball, the stationary ball is knocked backwards. Similarly, if a WIMP collides with the nucleus of an atom, the atom will recoil. The energy of a recoiling atom can be detected in various ways.

For example, in a semiconductor such as silicon, an atom recoiling from a dark matter collision would release free electric charge (ionisation) in the material (see Figure 2a). This can be collected and measured with an electronic circuit. Another method depends on an effect called 'scintillation' - in some crystals or liquids the recoiling atom causes the emission of a weak but measurable flash of light (see Figure 2b), and it is possible to detect such an event even if only a few photons of light are released. A third method exploits the fact that in a crystal, if a moving atom slows down, it will lose its energy in the form of vibrations or 'phonons' (see Figure 2c) which can be detected at temperatures close to the absolute zero. All of these methods have been clearly demonstrated in tests using known particles such as photons and neutrons.

It might, therefore, seem straightforward to detect dark matter in the form of WIMPs. In reality it turns out to be very difficult, because while we are looking for typically one collision per day in a one-kilogram block of material, that same material will be experiencing millions of collisions per day from other atomic particles in our environment - notably cosmic rays from space and gamma rays and neutrons from radioactivity in surrounding materials. This 'background' radiation also produces signals in the material by collisions with atoms. So any signal from WIMPs will be totally swamped by these background effects unless we take steps to remove them.

The apparatus can be isolated from background radioactivity by placing it in an enclosure shielded with, for example, pure lead or pure water. Cosmic rays cannot be excluded in this way because they include particles called muons (a heavy form of electron) which penetrate the shielding material and interact with it, knocking neutrons out of atomic nuclei. These neutrons will collide with nuclei in our detector, and we would be unable to distinguish these collisions from those caused by WIMPs.

There is only one certain way to remedy this - a course that particle physicists have adopted many times to shield their experiments from cosmic rays. The equipment must be taken deep underground. At depths of 1000 metres or more the muons in cosmic rays are almost entirely blocked by the overhead rock. The subterranean site is usually at the bottom of a deep mine or a road tunnel under a mountain.

About a dozen such sites are already used by physicists around the world. Examples of mountain sites are the Gran Sasso tunnel in Italy, the Frejus tunnel on the French-Swiss border and the Canfranc tunnel on the French-Spanish border. Examples of experiments in deep mines are to be found at the Homestake gold mine in South Dakota, the Soudan iron mine in Minnesota and the Creighton nickel mine at Sudbury, Canada. Some of these experiments are set up to detect neutrinos from the Sun and other stars. Others are looking for rare decays of atomic nuclei or of the proton itself. These experiments could detect some types of WIMP, but their sensitivity is limited, and new detectors need to be developed.

When British scientists first started to consider carrying out dark matter experiments, they thought they would have to use one of the underground sites in the US or Europe. However, the British team were surprised to learn that Britain has the deepest mine in Europe. This is a working salt and potash mine run by Cleveland Potash at Boulby, on the northeast coast of England near Whitby. It consists of a large network of tunnels and caverns at a depth of 1100 metres. What is more, the salt rock is particularly low in radioactive impurities, making it one of the best sites in the world for experiments requiring low levels of background radiation.

Several years ago, when the British collaboration approached the managers of the Boulby mine, they showed an immediate interest in the project. They allowed the UK team into the mine to confirm the low radiation levels in the caverns, and then provided a site for dark matter experiments by enlarging an existing cavern and laying on electrical power, lighting and ventilation. A dustproof cavern liner, telephone lines, and a data communications cable to the surface were added later.

In 1990, the project received full funding, and the next step was to eliminate the effects of the natural radioactivity in the cavern walls. For example, any uranium present emits photons (X-rays and gamma rays) which would be registered by a detector. The levels of natural uranium in the Boulby mine are extremely low - less than one part in 10 million - but even this would still be sufficient to spoil a dark matter search. The solution adopted was to install a shielding system consisting of 200 tonnes of high-purity water, in which the experiments can be suspended in waterproof containers.

There remains the formidable task of designing detectors that are sensitive enough to detect a few collision events per day. Detectors are being developed based on the ideas shown in Figure 2, but the most serious problems lie with the purity of the materials from which they are made. Even with the best materials available radioactivity in the detector itself could mask any signal from WIMPs. Because of this the researchers are attempting the experiments in several stages. Theorists do not know the precise rate of dark matter collisions but estimate that it could be as many as 100 events per day per kilogram of material, or as few as one event per day in 100 kilograms. So as a first step the British team is trying to construct a detector sensitive enough to pick up between 10 and 100 events a day.

Design for a detector

They are now assembling a detector that may achieve this. It is a scintillation detector containing a crystal of pure sodium iodide, about 7 centimetres in diameter and weighing 1 kilogram. Particles that interact with the crystal release bursts of light which can be detected by standard photomultipliers. Because these photomultipliers can detect single photons of light they can register very low energy events in the crystal. However, the glass windows and casing of the photomultipliers contain radioactive impurities, so they must be shielded from the crystal by blocks of copper or lead. The light then has to be channelled to the photomultipliers by mirrors made of aluminium foil. Working out how to arrange the component parts of the system proved to be a difficult design problem, and required specially written computer programs to arrive at the best solution. The whole system will be enclosed in a watertight copper container and tested in the underground tank in June this year - possibly ahead of Italian and French groups who are developing similar experiments. These first tests should reveal whether the crystal is pure enough to detect fewer than 100 events per day.

Preparing and assembling even this simplest of experiments to detect dark matter requires meticulous care. As well as needing pure materials, an unusual problem arises from radon gas, which exists everywhere in the air (being a product of the decay of uranium in the ground) and decays in three days to a radioactive form of lead which becomes deposited on surfaces. To avoid this contamination all surfaces are cleaned and assembled in a tent filled with 'old air' - air from a diving cylinder several weeks old. The experiment is then enclosed in a radon-proof bag to avoid being contaminated further while being transported to the mine and tested underground.

Because there are uncertainties about the purity of crystal scintillators, the British team is also developing other types of detector. One option exploits scintillating liquids which have a higher purity. Another is based on gallium arsenide (see Figure 2). However, it is unlikely that these 'stage-one' experiments will detect rates below 30 events per day.

The way to improve on this is to develop a second-stage detector that can identify which events are genuine nuclear collisions with WIMPs and which events are due to background radiation. This remarkable possibility can be achieved in several ways. One way is to combine two different detectors into a hybrid instrument that can measure, for example, both ionisation and phonon signals. Both the British team and a rival American group have shown that the ratio of the two signals resulting from nuclear collisions is quite different from the ratio when the signals are caused by radioactive impurities. The 'shape' of the phonon signal itself might also differ for collisions by dark matter particles. Another possibility being studied uses liquid xenon as a target. This gives both scintillation and ionisation signals, and again the ratio of the signals for dark matter collisions would be different from the signals for other sources. Moreover, liquid xenon has relatively few impurities.

The ideas behind both of these 'stage-two' experiments are also being developed by rival research groups. The first is the basis of a dark matter experiment planned by researchers in the US , while the second is being developed by researchers at the University of Rome for a possible experiment in the Gran Sasso Laboratory.

If signals are seen, how will we be certain that they are caused by dark matter particles? One way of finding out would be to repeat the experiments using detectors made of different materials. By measuring the collision energy with different target atoms we could estimate the mass and properties of the dark matter particles. But the clearest proof would come from the annual modulation. Because the Earth revolves around the Sun and the Sun moves through the Galaxy, the speed of the Earth through the Galaxy depends slightly on the time of year. This means that the rate and energy of collisions between dark matter particles and the detectors would also vary slightly. Calculations show that we expect a signal in June that is about 10 per cent greater than the signal in December.

In principle, an even bigger effect should result from the direction of the signal. Dark matter particles should have velocities of about one-thousandth of the speed of light, and would be moving in all directions in the Galaxy. But because we are moving through the galaxy at about the same speed, more of the particles - and therefore more collisions - would appear to be coming from the direction in which the Earth and Sun are moving. This effect would be quite large, and would show clearly that our signal is caused by something in the Galaxy.

Unfortunately, none of the detectors planned so far can measure the direction of the particles. Although the target atoms are indeed knocked backwards in a specific direction, they move only about one millionth of a centimetre in the material - too small to measure the direction.

Finally there is still the possibility that dark matter might consist of normal matter in a form which does not emit light - perhaps large numbers of black holes or dead stars. Two research groups, in France and the US, are now starting to look for such objects. Their aim is to continuously monitor about a million stars in the cluster of stars known as the Large Magellanic Cloud close to our own Galaxy. If a large non-luminous object in our galaxy were to pass between us and any of the stars being monitored, its gravity would act as a lens, concentrating the starlight to produce an increase in brightness. If the Galactic dark matter is really due to large objects with mass less than a normal star one should see typically 10 to 100 of these brightening events a year. But if nothing is seen this will further intensify the search for new particles as the most likely explanation of the dark matter in the Universe.

Peter F. Smith is a member of the Particle Physics Department, Rutherford Appleton Laboratory, Didcot, Oxfordshire.

* * *

How the race was begun

The identification of dark matter has been designated by the particle physics community in Britain, in its report 'Particle Physics 2000', as one of the priority objectives for this decade. The Science and Engineering Research Council approved funding for experiments in 1990, shared equally between the Council's budgets for particle physics and astronomy. But the theoretical study of dark matter goes back much further.

Astronomers and cosmologists in Britain and elsewhere have been investigating dark matter since the late 1970s. Important British contributions came from the universities of Cambridge, Sussex, Oxford, Durham and from Queen Mary and Westfield College, London. Theoretical astronomers have also helped to promote the first proposals for dark matter experiments.

Studies of possible practical experiments were carried out in the early 1980s at the Rutherford Appleton Laboratory. In 1986 this expanded into an interdisciplinary collaboration of all interested experimental physicists in Britain. Most of the participants are based at two centres: the Particle Physics Department at RAL and the Astrophysics Group of Imperial College London. Others involved are the Solid State and Particle Physics departments at Imperial College, low temperature groups at Royal Holloway and Bedford New College and at the University of Nottingham, and cosmic ray physicists at Nottingham and Birkbeck College.

Worldwide, there are now a dozen research groups developing dark matter detectors, with three or four front-runners in the race to set up the first experiments this year. But, as in any race, there are tactics. The British and European groups are currently competing to set up stage-one experiments, with more advanced stage-two experiments to follow. But the main American effort in this field, based at the Center for Particle Physics at Berkeley in California, hopes to jump ahead by going straight for a more difficult but more sensitive stage-two experiment. Because of this, the British collaboration also has a strong stage-two programme, now funded until 1995.

The British team believes that it has an important advantage in the low background facility installed in the Boulby mine. Progress now depends on the speed with which the rival groups can put their ideas into operation. We can expect new levels of sensitivity to be reported within the next 12 months - perhaps by several groups. Realistically, however, it could take at least another one or two years before we reach the levels of sensitivity which might detect dark matter for the first time.


From New Scientist magazine, vol 134 issue 1818, 25/04/1992, page 37