Compensating Calorimeter

  In hadronic and combined electromagnetic/hadronic calorimeters, the energy resolution achievable for hadrons is critically dependent on the choice of absorber and active materials, and their relative thicknesses. It is important that the energy response at all energies is as independent as possible of the fluctuations in shower development, in particular the content of electromagnetic particles (electrons and gammas). This is of prime relevance for the measurement of jet energies, as in this case not only electromagnetic particles may appear during shower development, but the content (and hence the fraction of energy in the form of 's) can be substantial in the jet before it impinges on the calorimeter.

In general, the average ratio between signals from electromagnetic and hadronic particles of the same incident energy is calorimeter- and energy-dependent, and for non-compensating calorimeters there is a higher response for electromagnetic particles, typically

For a compensating calorimeter, the electron/hadron signal ratio should be close to one.

Various phenomena in both active and passive layers of sampling calorimeters can be put to use to achieve , thus optimizing energy resolution: adjusting the relative thickness of absorber and active layers, using U²³⁸ as absorber for its fission capability for slow neutrons, or shielding the active layers by thin sheets of low-Z material to suppress contributions from soft photons in electromagnetic showers, are possible methods of active compensation (see [Wigmans91a]).

The photon absorption in the (high-Z) absorber material plays a significant role, and so does the conversion of low-energy neutrons into signal, e.g. by detection of de-excitation photons; the hydrogen content in the active medium is relevant here. For a detailed discussion, see [Wigmans91b].

If high resolution is not required during readout, e.g. for triggering, corrections corresponding to compensation may also be applied by an a posteriori algorithm (``off-line''), when the shower profile (mostly the longitudinal distribution) is known ( [Fesefeldt90a], [Andrieu93]). Just how much can be recovered by calibrations of this type, is strongly detector-dependent; [Borders94] has explored the possibilities for a specific non-compensating sampling calorimeter in detail, using individual weights for sampling layers.