Le sang froid du CERN

Quand le CERN lancera l'exploitation du LHC, il s'appuiera sur les plus grands systèmes mondiaux de réfrigération et de distribution d'hélium à 1,8 K, tandis que les collaborations ATLAS et CMS feront appel à une palette impressionnante de techniques cryogéniques. Cependant, l'utilisation de la cryogénie au CERN remonte à l'époque de certaines des premières expériences, d'abord dans les détecteurs et ensuite autour des accélérateurs.

Cryogenics at CERN has now reached an unprecedented scale. When the Large Hadron Collider (LHC) starts up it will operate the largest 1.8 K helium refrigeration and distribution systems in the world, and the two biggest experiments, ATLAS and CMS, will deploy an impressive range of cryogenic techniques. However, the use of cryogenics at CERN, first in detection techniques and later in applications for accelerators, dates back to some of the earliest experiments.

The need for cryogenics at CERN began in the 1960s with the demand for track-sensitive targets - bubble chambers - that contained up to 35 m3 of liquid hydrogen, deuterium or neon/hydrogen mixtures. These devices required cryogenic systems on an industrial scale to cool down to a temperature of 20 K. For more than a decade they were a major part of CERN's experimental physics programme (see CERN Courier July/August 2004 p26). At the same time, cryogenic non-sensitive targets were used in other experiments. Over the past 30 years some 120 such targets have been constructed, ranging in size from a few cubic centimetres to about 30 m3 and usually filled with liquid hydrogen or deuterium, again requiring cooling to 20 K.

Cool targets, cool detectors

At the smallest scale, the demand from the fixed-target programme for polarized targets at very low temperatures led to the development of dilution refrigerators at CERN in the 1970s (figure 1). Going below the range of helium-3 evaporating systems, these require small-scale but highly sophisticated cryogenic techniques.

Polarized targets remain very much part of the current physics programme at CERN, where the COMPASS experiment uses solid targets made of ammonia or lithium deuteride. The basic method for obtaining a high polarization of the nuclear spins in the targets is the dynamic nuclear polarization process. This uses microwave irradiation to transfer to the nuclei the almost complete polarization of electrons that occurs at low temperatures (less than 1 K) and in a high magnetic field (2.5 T), generated by a superconducting solenoid.

On a larger scale, in detector technology the development in the 1970s of sampling ionization chambers - calorimeters - broadened the demand for low temperatures at CERN. Using liquid argon to measure the energy of ionizing particles, these detectors required cryogenic systems to cool down to 80 K. Several calorimeters, with typical volumes of 2-4 m3, were built in this period, both for fixed-target experiments and use at CERN's first collider, the Intersecting Storage Rings (ISR) - which was also the world's first proton collider.

Two decades later, in 1997, the NA48 experiment extended the technique from argon to krypton. With its very high density, liquid krypton not only provides the "read out" through the ionization of the liquid by charged particles, but also acts as a passive particle absorber, so avoiding the use of a material such as lead or uranium. The cooling fluid for this detector is saturated with liquid nitrogen and the heat is extracted by re-condensing the evaporated krypton via an intermediate bath of liquid argon, which in turn feeds the 10 m3 liquid-krypton cryostat by gravity.

Around the same time as the development of the first liquid-argon calorimeters, experiments began to require helium cryogenics, mainly at 4.5 K, for superconducting magnets. These were used to analyse particle momenta in magnetic spectrometers. The largest built for the fixed-target programme at CERN was the superconducting solenoid constructed for the Big European Bubble Chamber (BEBC) in the 1970s (figure 2). This had an internal diameter of 4.7 m and produced a field of 3.5 T. The associated combined He/H2 refrigeration system had a cooling capacity of 6.7 kW at 4.5 K.

With the advent of the Large Electron-Positron (LEP) collider at the end of the 1980s, collider experiments took on a much greater role at CERN. Two of the LEP experiments, ALEPH and DELPHI, opted for large superconducting solenoids for momentum analysis - the choice between superconducting and normal (resistive) magnets depending on considerations related to "transparency" (to particles) and/or economy. Each of these solenoids required a helium cooling system of 800 W at 4.5 K.

A current novel application for a superconducting magnet occurs in the CAST experiment located on the surface above the cavern where the DELPHI experiment for LEP was installed. This uses a 10 m, 9.5 T prototype LHC superconducting dipole and also makes use of the DELPHI refrigerator to cool the superfluid helium cryogenic system for the magnet. The aim of the experiment is to detect axions, a possible candidate particle for dark matter that could be emitted by the Sun, through their production of photons in the dipole's magnetic field (see CERN Courier March 2005 p7).

Now, however, the major effort at CERN is focused on the LHC, with four big experiments: ALICE, ATLAS, CMS and LHCb. Basic design criteria led the two largest experiments, ATLAS and CMS, to construct superconducting spectrometers of unprecedented size, while ALICE and LHCb opted for resistive magnets.

ATLAS has several components for its magnetic spectrometry. A "slim" central solenoid (with a length of 5.3 m, a 2.4 m inner diameter and a 2 T field) is surrounded by a toroid consisting of three separate parts - a barrel and two end-caps. The overall length of the toroid is 26 m, with an external diameter of 20 m (figure 3). It is powered up to 20 kA and has a stored energy of 1.7 GJ. CMS, by contrast, is built around a single large solenoid, 13 m long, with an inner diameter of 5.9 m and a uniform field of 4 T (figure 4). When powered up to 20 kA it has a stored energy of 2.6 GJ (see CERN Courier March 2004 p6).

ATLAS also has a cryogenic electromagnetic calorimeter, with the largest liquid-argon ionization detector in the world to measure the energy of electrons and photons (see CERN Courier September 2005 cover and p7). This consists of a cylindrical structure made of a barrel and two end-caps, with a length of 13 m and an external diameter of 9 m. Altogether, the cryostats for the three sections contain 83 m3 of liquid argon and operate at 87 K.

Both ATLAS and CMS have refrigerating plants that are independent from the system required to cool the LHC to 1.8 K (see below). ATLAS will use two helium refrigerators and one nitrogen refrigerator, while CMS will have a single helium refrigerator. These will provide cooling for current leads and thermal shields, as well as for the refrigeration at 4.5 K for the spectrometer magnets, and in the case of ATLAS also at 84 K for the electromagnetic calorimeter.

Cool accelerators

The use of helium cryogenics was extended to accelerator technology at CERN during the 1970s, when superconducting radiofrequency beam separators were constructed for the Super Proton Synchrotron, and superconducting high-luminosity insertion quadrupoles were built for use at the ISR. These required cooling of 300 W at 1.8 K and 1.2 kW at 4.5 K, respectively. The 1990s saw the larger scale use of cryogenics for accelerators with the upgrade of LEP to higher energies.

LEP was built initially with conventional copper accelerating cavities, but with the successful development of 350 MHz superconducting cavities in 1980s, its energy could be doubled. As many as 288 superconducting cavities were eventually installed, increasing the energy from 45 to 104 GeV per beam (figure 5). This involved the installation of the first very large capacity helium refrigerating plant at CERN, with four units each of a capacity of 12 kW at 4.5 K, later upgraded to 18 kW, supplying helium to eight 250 m long strings of superconducting cavities, and a total helium inventory of 9.6 tonnes.

LEP was closed down at the end of 2000 to make way for the construction of the LHC in the same tunnel. This liberated most of the existing cryogenic infrastructure from LEP for further use and upgrading for the LHC, which will require the largest 1.8 K refrigeration and distribution system in the world to cool some 1800 superconducting magnet systems distributed around the 27 km long tunnel (see CERN Courier January/February 2004 p27). A total of 37,500 tonnes has to be cooled to 1.9 K, requiring about 96 tonnes of helium, two-thirds of which is used for filling the magnets.

Although normal liquid helium at 4.5 K would be able to cool the magnets so that they become superconducting, the LHC will use superfluid helium at the lower temperature of 1.8 K to improve the performance of the magnets. The magnets are cooled by making use of the very efficient heat-transfer properties of superfluid helium, and kilowatts of refrigeration power are transported over more than 3 km with a temperature difference of less than 0.1 K.

The LHC is divided into eight sectors, and each will be cooled by a two-stage cryoplant consisting of a 4.5 K refrigerator coupled to a 1.8 K refrigeration unit. The transport of the refrigeration capacity along each sector is made by a cryogenic distribution line, which feeds the machine every 107 m. A cryogenic interconnection box will link the 4.5 K and 1.8 K refrigerators and the distribution line. Together the refrigerators will provide a total cooling power of 144 kW at 4.5 K and 20 kW at 1.8 K. The 4.5 K refrigerators are equipped with a 600 kW liquid-nitrogen precooler, which will be used to cool down the corresponding LHC sector to 80 K in less than 10 days.

Four new 4.5 K refrigerators built by two industrial companies have been in place since the end of 2003 (see CERN Courier May 2004 p15), and four 4.5 K refrigerators recovered from LEP are being upgraded for use at the LHC. In addition, eight 1.8 K refrigerator units procured from industry provide the final stage of cooling (figures 6 and 7). Four 1.8 K units built by one company have already been installed; the other four units, made by the other company, are currently being installed and will be tested in 2006.

For the next 15 years or so, CERN will need to continue to provide strong support in cryogenics for its unique accelerator facilities, including the final consolidation and operation of the LHC. Further long-term perspectives will depend a great deal on the next generation of accelerators. Detectors, on the other hand, have proved quantitatively less demanding for cryogenics in comparison with the accelerators; however, over the years their cryogenic needs have generated a variety of different applications, with a temperature range from 130 K (liquid-krypton calorimeters) down to a few tenths of a millikelvin for polarized targets. Innovation in detector technology has often in the past led to the application of cryogenics - a trend that will no doubt continue into the future.

• This article is based on: G Passardi and L Tavian 2002 Cryogenics at CERN Proceedings of the 19th International Cryogenic Engineering Conference (ICEC 19); L Tavian 2005 Latest developments in cryogenics at CERN Proceedings of the 20th National Symposium on Cryogenics, Mumbai (TNSC 20).