The ALICE experiment is optimized to perform in the environment of heavy-ion collisions at the LHC, which can produce thousands of particles. Its design combines an excellent vertex resolution with a minimal thickness of material. It has excellent performance for particle identification in a large range of momenta as it employs essentially all of the known relevant techniques. Accurate knowledge of the geometry and chemical composition of the detectors is particularly important for tracking charged particles, for the calculation of energy loss and efficiency corrections, as well as for various physics analyses such as those involving the antiproton–proton ratio, direct photons and electrons from semileptonic decays of heavy-flavour hadrons.

The γ-rays produced in proton–proton collisions at the LHC (mainly from π0 decays), which undergo pair production in the ALICE experiment, provide a precise 3D image of the detector, including the inaccessible innermost parts. The process is almost exactly the same as in 1895 when Wilhelm Röntgen produced an X-ray image of his wife’s hand – the inner parts of the body could be seen for the first time without surgery. The main difference lies in the energy of the radiation – of the order of 100 keV for Röntgen’s X-rays compared with more than 1.02 MeV for the γ-rays from pair conversions. Importantly for the ALICE experiment, it allows the implementation of the detector geometry in GEANT Monte Carlo simulations to be checked.

To produce the γ-ray image, photons from pair conversions are reconstructed through the tracking of electron–positron pairs using a secondary vertex algorithm. Contamination from other secondary particles, such as K0S, Λ and Λ, is reduced by exploiting ALICE’s excellent capabilities for particle identification. In this case, the analysis uses the specific energy-loss signal in the time-projection chamber (TPC) as well as the signal in the time-of-flight (TOF)detector. Photons from pair conversions provide an accurate position for the conversion vertex, directional information and a momentum resolution given by that for the charged particles – an advantage over calorimeter measurements at low transverse momentum.

Figure 1 shows the γ-ray picture of the ALICE experiment, i.e. the Y-distribution versus X-distribution of the reconstructed photon conversion vertices, compared with the actual arrangement used in the 2012 run. The different layers of the inner tracking system and the TPC, as well as their individual components (ladders, thermal shields, vessels, rods and drift gas), are clearly visible up to a radius of 180 cm. To obtain a quantitative comparison, the radial distribution of reconstructed photon conversion vertices normalized by the number of charged particles in the acceptance is plotted together with the Monte Carlo distributions in figure 2.

This indicates an excellent knowledge of the material thickness of the ALICE experiment: up to a radius of 180 cm and in the pseudorapidity region |η| < 0.9, the thickness is 11.4±0.5(sys.)% of a radiation length. The systematic error is obtained from a quantitative comparison of the data with the Monte Carlo distributions, after taking into account the limited knowledge of the true photon sample, of the photon reconstruction efficiency and of the geometry and chemical composition of the detectors.

The accuracy achieved, as well as the full azimuthal acceptance of the central barrel, allows converted photons to be used in physics analyses. So far, photons from pair conversions have been used for the identification of neutral mesons in proton–proton collisions at 7 TeV down to a transverse momentum of 0.3 GeV/c – the first time in a collider experiment. Moreover, a direct photon signal observed in lead–lead collisions at √sNN = 2.76 TeV has been measured with the photon-conversion method. The latter measurement demonstrates that the quark–gluon plasma formed at the LHC is the hottest matter ever made in the laboratory (CERN Courier December 2012 p6).