The investigation of the properties of nuclear matter at high densities and high temperatures is one of the main goals of experiments with relativistic heavy ions. In experiments carried out at over 100GeV per nucleon (CERN SPS), the highest ion energy currently available, the energy densities reached in heavy ion collisions are such that a phase transition from a hadron system to a quark-gluon plasma is expected. A considerable modification of the properties of the hadrons is already expected at much lower ion energies due to the increased density inside the reaction zone. In order to investigate this phenomenon at energies of a few GeV per nucleon, an international collaboration was formed to set up HADES, a completely new type of di-lepton spectrometer, at the GSI.
Figure 1: shows the effective mass of a constituent quark as a function of the baryon density and temperature. The mass mH of the hadrons (with the exception of the pion) should follow the path shown in the figure . The shaded areas mark the densities and temperatures which can be achieved in heavy ion collisions for ion energies which can be reached at the SIS (1-2 GeV per nucleon) and at CERN SPS (160-200 GeV per nucleon). Additionally, one sees that observable deviations of the particle masses are already to be expected at normal nuclear densities and should be observable by means of pion induced reactions.
In the experiments to investigate high density nuclear matter which have been carried out so far at GSI, two main approaches have usually been followed. One main focus of the research was the dynamics of the hot dense reaction zone created in the heavy ion collisions, the other was the production of new particles, particularly mesons, in the high density phase of the reaction (see GSI-Nachrichten 6/96 and 1/97). In the future, another important aspect, the direct determination of the effective hadron masses and the coupling constants in compressed nuclear matter will be covered. According to theoretical predictions a partial restoration of chiral symmetry-a fundamental symmetry of the strong force-which is broken under normal circumstances, is expected at increased compression. As a result, the effective mass of the quarks, and therefore also the mass of the hadrons should be reduced. In research on this phenomenon, spectroscopy of the vector mesons , and will be of foremost importance. Of these, the -meson and to some extent the -meson decay within the compressed collision zone due to their short half-lives. The electromagnetic decay of the vector mesons into an electron and a positron allows the direct determination of the masses through the reconstruction of the momenta. The decisive advantage of this method is that the leptons, in contrast to hadron probes, can leave the reaction zone undisturbed. Final state interactions take place a factor of 2=1/19000 less often. However, this advantage is coupled to the problem that the probability of decay into a lepton pair (di-lepton) is reduced by about the same factor. This reduction factor makes particle spectroscopy using this decay channel more difficult and places correspondingly higher requirements on the experimental set-up.
In addition to the decay of short lived vector mesons there are a number of continuum sources for leptons such as the Dalitz decay of the (, ) and bremsstrahlung processes. The latter allow conclusions to be made about the dynamics of the reaction. The Dalitz-decay of the short lived can be used to a certain degree for the spectroscopy of these resonances in nucleus-nucleus collisions.
The small probability of the particles decaying into a di-lepton discussed above defines the boundary conditions which an experiment on the spectroscopy of di-leptons must take into account. As far as possible, the leptons produced must be reliably detected within the total solid angle. At the same time the error rate for the identification of the leptons may not be larger than one in ten thousand. The momenta of the leptons must be measured with a good resolution (about 1 percent) and simultaneously over a wide momentum range (100 MeV/c to 1000 MeV/c). In this way not only is it possible to reconstruct the masses of the decaying hadrons-for example in order to be able to separate the -meson from its isospin partner the -meson-but also their momentum relative to the center-of-mass system can be efficiently measured at the same time. Good momentum resolution is also required to separate the decays of free particles outside the dense nuclear matter zone from those inside it. Another important point is the processing of high reaction rates of about a million events per second, in order to carry out the spectroscopy with the necessary statistics.
In addition to the small probability of decay into di-lepton pairs there is the further problem of an intensive lepton source which dominates the di-lepton spectrum even if lepton identification is perfect. This is the Dalitz decay of the neutral pion (0). As the decay channels of the 0, the lightest meson, are exclusively electromagnetic, the Dalitz decay at about 1.3 percent is the dominant source of electrons. As soon as several 0's are produced in the same reaction, an electron from one 0-decay can be combined with the positron of another 0-decay, so that the reconstructed masses create a continuous spectrum. In the case of heavy ion collisions this "combinatoric" background dominates even for large invariant masses in the vector meson region (800 MeV/c2) of the measured spectrum. The 0-decay into a positron, electron, and a gamma does however differ in a characteristic way from the two body decay of interest. The opening angle between the two leptons is generally smaller. However, in order to benefit from this effect, both leptons of the majority of 0-decays must be detected.
In order to successfully investigate lepton decay as outlined above, a dedicated detector system is required. The unequivocal identification of lepton pairs with good mass resolution and at high reaction rates required the development of a new type of detection system for leptons and, in particular, new types of processing electronics and trigger systems. The HADES collaboration was founded in 1994 to carry out this development. HADES stands for "High Acceptance Di-Electron Spectrometer". The collaboration currently includes over a hundred members from 18 institutions, located in eight European countries. Some of the developments carried out by the collaboration are presented briefly below.
Figure 2: Cross-section through the HADES spectrometer. The HADES-spectrometer consists of a "Ring Imaging Cherenkov Detector" (RICH) for the identification of leptons and a superconducting torus magnet surrounded by a total of 24 drift chambers (MDC) for the measurement of the deflection of charged particles in the magnetic field. An array of scintillators for time of flight (TOF) measurements, as well as a set of shower detectors, are used in addition for the identification of charged particles.
In order to optimize the design of the detector system-also with a view to cost-extensive simulation calculations were carried out. These were based on theoretical models describing heavy ion collisions, which were developed by theory groups of the Universities of Frankfurt and Gießen. The results of these model calculations were fine tuned using known measurement results to ensure maximum correspondence to "real life." The scenario of reaction products determined in this way was used to determine the optimum configuration of the detector system with respect to the detection probability, the identification of lepton pairs, and the cost of the system. A computational effort of several thousand hours on the fastest workstations available at GSI finally produced the concept described below .
The heart of the detector system is a so-called "Ring Imaging Cherenkov Detector" (RICH), whose job is to filter out leptons from the background of heavier charged particles (hadrons), which is many orders of magnitude higher. It is, so to speak, "blind" to hadrons. This is achieved by sending the particles through a gas chamber filled with C4F10 (Figure 2). Only very light particles which are moving at almost the speed of light, radiate UV-light (Cherenkov photons) in this gas. Hadrons are too slow, because of their large mass, to radiate Cherenkov photons. The light emitted from the leptons is deflected by a mirror, so that a ring of characteristic size is produced on a reverse mounted photon detector. The position of the ring on the detector depends only on the angle at which the lepton was emitted. This system is a technical challenge in two respects.
Figure 3: The detection probability for UV photons as a function of wavelength (l), for various CsI-layers. The layers were produced by different institutes using a variety of processes. The layer produced at the TU Munich (dotted line) reaches a remarkably high photon yield.The detector which measures the photons has to have a high efficiency and a good spatial resolution and must, at the same time, be extremely fast. At present this is only possible with gas counters which use a layer of cesium iodide (CsI) as a photocathode. Figure 3 shows the measured efficiency of CsI layers produced by different institutes using a variety of techniques. A prototype of the size required for the experiment achieved a detection probability for photoelectrons of over 95 percent. This is sufficient, in addition to the high conversion rate of photons into photoelectrons, to effectively recognise the rings as they occur. On average about 15 photons per ring are detected.
A further technical challenge was the RICH mirror. On the one hand it had to be very thin, despite its large area of 1.6 square meters, as all emitted particles must pass through it without this resulting in a significant loss of information. On the other hand it must reflect UV-light down to wavelengths of about 140 nanometers with high efficiency. This requires a surface roughness of about 1 nanometer (a millionth of a millimeter). In co-operation with DASA a process was developed for manufacturing special carbon layers of high stability and surface quality. The first samples show that the necessary surface quality will be achieved. The additional development costs could at least be partially compensated for, as this process dispenses with the need for an expensive glass mould for mirror production.
The second, important element of the system is a superconducting torus magnet, consisting of six coils, which deflects the charged particles. Drift chambers (MDC) in front of and behind the magnet allow the angle of deflection to be measured and thus the momenta of all charged particles to be determined. The mass of the hadron, which decayed previously in the nucleus, can be determined from the momenta of both partners of a lepton pair and the opening angel between the partners. Assuming, of course, that no third decay partner (e.g. a high energy gamma) was involved. Decisive for the choice of a toroidal magnetic field is the requirement that the emission of Cherenkov photons must occur in an area in which there is almost no magnetic field, otherwise the focusing of the photons to rings would be disturbed.
The superconducting torus magnet was ordered as a complete system from Oxford Instruments. It is noteworthy for its compact construction and low helium consumption. The six coils, through each of which a current of about 3,500 amps flows, have a width of eight centimeters including thermal isolation, support structure, and vacuum housing and consist, in addition to the Nb(Ti) superconductor itself, almost entirely of aluminum. A force of 24 tons acts on the coils at the maximum magnetic field of 3.7 Tesla.
The simulations showed that, in order to achieve a reduction of hadrons to the level of 1:10000, it is not enough to identify the leptons only once. Therefore this system is surrounded by an array of scintillators which are used to measure the time of flight of the charged particles. Additionally, a set of shower detectors which are mounted in the forward direction allow a further independent identification of the leptons.
The system has sixfold symmetry and allows particle detection at all azimuth angles (with the exception of the shadow of the six coil-cases of the torus magnet). The polar angle region between 18° and 85° relative to the beam axis is covered. About 40 percent of all lepton pairs which are created in the heavy ion collisions at SIS energies are emitted in this region.
Once the experimental set up was decided on, the quality of the measurement data which can be expected could be checked in simulation calculations. The results of the simulation are summarised in Figure 4. The reconstructed mass distribution for the known decay of hadrons into leptons is shown. An expected, but theoretically difficult to quantify change in the particle masses in dense, hot nuclear matter was not taken into account in this figure. However, it can be seen that the resolution of the spectrometer should be good enough to measure possible changes in the region of several 10 percent in the mass of decaying hadrons. The relative contributions of the individual decay channels vary strongly with the injection energy of the ion beam. In this case an energy of 1 GeV per nucleon was assumed. A more detailed discussion of the expected change in the spectral distribution of the reconstructed particle masses for different collision systems and injection energies would go beyond the scope of this article. An interesting special case for reactions induced by pions is described in . The detector properties which were incorporated in the analysis have in the meantime been verified by measurements on prototypes.
Figure 4: Reconstructed mass distribution for lepton pairs from the collisions of two gold nuclei at an injection energy of 1 GeV per nucleon. The detector properties on which the simulation was based were verified on prototypes. Possible changes in the particle masses in dense, hot matter were not taken into consideration.
Prototypes of all essential detector components have already been subjected to extensive tests. Important milestones towards to the goal of bringing the HADES spectrometer into operation have been set: in July of this year a system test will be carried out in which all components in one sector of the complete spectrometer will be operated together. In addition to the detector components-RICH, drift chamber, time-of-flight detector and shower detector-a prototype of the processing electronics will be used for the first time. In the first half of next year the beam line, the cave and counting room, and the necessary infrastructure at the cave and counting room will be set up. Then in the summer of 1998 the superconducting magnet will be installed and brought into operation. The final step will be mounting the various detectors. The first test of the complete spectrometer is planned for the end of 1998.
 R. Schicker et al: Nucl. Inst. Meth. A330 (1996) 586
 Nuclear Physics A 610 (1996) 342c
 Act. Phys. Pol. B27 (1996)