The PANDA experiment at FAIR

The Edinburgh Nuclear Physics group has been participating since several years in the PANDA activities. At the upcoming FAIR laboratory the PANDA collaboration prepares to build the PANDA experiment, a fixed target experiment scattering cooled antiprotons circulating in the High Energy Storage Ring with momenta of 1-15 GeV/c off an internal pellet or gas jet target at interaction rates of up to 2x10E7 1/s to perform high precision experiments in the charmed quark sector.

For the charged particle identification (PID), in particular of kaons, two DIRC detectors are foreseen as dedicated detectors in the PANDA target spectrometer, the leftmost part in the figure below.

This target spectrometer is almost hermetically sealed to avoid solid angle gaps, and to keep material volume low, there is little spare space inside. The possibility of using thin radiators and placing the readout elements outside the acceptance, favours the use of DIRC designs as Cherenkov imaging detectors for PID.

PANDA at FAIR (visualisation from 2005)

For the forward Endcap inside the target spectrometer, we from Edinburgh have suggested a novel DIRC design with a circular solid state radiator equipped with optical imaging readout elements on the outside rim.

PANDA detector (geometrical planning status spring 2007)

The detector and in particular the amorphous fused silica, the optical radiator we plan to use, are being located in a high radiation area. As one of the results of a joint proton beam irradiation our Glasgow colleagues have tested several samples of commerially available fused silica material and found them to be sufficiently radiation hard.

The photon detectors have to stand high photon rates and a high magnetic field. Several universities from the PANDA Cherenkov group are testing different types of photon detector.

Focussing Lightguide Readout

We from Edinburgh have put forward a Focussing Lightguide Dispersion-Correcting design for an Endcap DIRC. When a photon arrives at the edge of the circular (or polygonal) disc, it enters into one of 100-200 optical elements on the rim. As we have to improve over currently implemented DIRC designs (like BaBar), we need to address the effects of chromatic dispersion of the Cherenkov light emission, and we have found that a prism element can largely correct the dispersion angle spread.     

In the lightguide side view the blue and red rays show how the LiF acting as a prism element aligns the initially wavelength-dependent angle of the Cherenkov radiation. The two boundary surfaces, with the radiator disc and the subsequent lightguide, make the chromatic dispersion correction angle-independent in first order.

The green lines show a set of rays used for optimising the lightguide curvature. Reflections at the parallel front and back surfaces keep the light inside but do not affect the focussing properties. The photons are focussed onto the readout plane shown in red. Here only the vertical coordinate is read out. While measuring the position on the red focal plane of the focussing lightguide yields the theta coordinate, the optical element entered determines the phi coordinate.

Pattern Recognition

In the figure to the left are photon patterns from four different particles. These shapes do not only depend on the particle velocity (that is the PID variable we are after) but also on the particle vertex. For the full PID, information from these shapes will be combined with adjacent tracking and calorimetry detectors.

The human eye can easily identify such a photon pattern from a particle, even if we were to add some noise to the the figure shown. With a simple pattern analysis algorithm using high level programming language on a standard computer we have estimated the detector performance analysing such hit patterns. But in the real experiment one has to be fast, and one has to use different (lower-level) hardware.

The PANDA DAQ and Computing group works to provide standardised front end Computing Nodes, based on high end FPGA circuits to process the continuously sampled data. Programming these beasts (and configuring the network connection topology) is quite different from standard computer programming. Front-end electronics provide time-stamped hit information as input, and several outputs are required:

  • separate physics channels in the subsequent analysis of the data
    (that could in principle be done offline as well)
  • fast PID information for online physics triggers
  • major data size reduction

Algorithms for this are not available off-the-shelf from existing detectors. Reasons are not only the different type of algorithms (for instance of the pattern recognition type) appropriate for FPGA, but also the novel Disc-DIRC design giving a new type of pattern.