The energy resolution of an electromagnetic calorimeter is usually written as the square root over the quadratic sum of three terms
The noise term a takes into account the electronic noise, pile-up and the radioactivity of the active and/or inactive medium. The term b is the square root of the quadratic sum of four terms, which are the sampling fluctuations, the fluctuations due to lateral escape, the Landau fluctuation and the intrinsic shower fluctuations due to the signal generating process. In noble liquid electromagnetic calorimeters, the contribution of the latter two are negligible. Landau fluctuation play an important role only in low density shower counters. Intrinsic fluctuations are negligible, since the W values of liquid argon and liquid krypton are typically seven orders of magnitude smaller when compared to interesting incident energies above 1 GeV. The quantity W is defined as the energy expended per ion pair. As a matter of fact, losses of roughly a factor of 50 can be tolerated in the signal generating process without interfering with the "stochastic" term b of the energy resolution. This article will outline a possible reduction of sampling fluctuation in exchange for tolerable losses in signal. The constant term c takes into account intercalibration errors between cells, inhomogeneities and energy absorbing material in front of the calorimeter. In a certain sense, c can be considered as the quality factor of a calorimeter. Constructional and mechanical errors will reflect themselves in a larger constant term. The stochastic term b and the constant term are often used to make comparisons between calorimeters without taking into account that calorimeters are built for specific experiments. These may require excellent time and space resolution in exchange for a slightly worsened energy resolution.
Calorimetry with noble liquids, like liquid argon or liquid krypton exhibit a
certain number of advantages: their response is uniform, various electrode
geometries are possible, they allow direct calibration, since the signal
generating process is proportional to the incident energy. In addition,
they are fast if limited to the measurement of the initial current. However,
they may be difficult to build, not only because of associated cryogenics.
