The CRESST experiment searches directly for
dark matter particles via
their elastic scattering off nuclei. The nuclei are in the absorber of a
cryogenic detector, capable of detecting the small energy of the recoiling nucleus which has been hit by an incoming dark matter particle.
Such a search for very rare interactions with a low energy
deposit involves two major aspects: a sensitive detector
and a highly efficient suppression of backgrounds (since also many other particles apart from dark matter particles interact in the detector and have to be discriminated from
the interesting signals).
Cryogenic Calorimeters
A cryogenic calorimeter consists of an absorber and a temperature sensor in
thermal contact, weakly linked to a heat bath.
In an extremely simplified model the detector can be characterized
as an absorber with a heat capacity C. Then an energy deposition in the absorber δE
leads to a temperature rise δT of the detector
given by δT=δE/C. This relaxes back to its
equilibrium value via the thermal coupling to the heat
bath. The temperature rise is therefore a direct measurement of the
deposited energy.
In dielectric and semiconductor materials the heat capacity at low
temperatures is dominated by the phonon system in which C ∝ T3. At
millikelvin temperatures, due to the T3
dependence of the heat capacity, the energy deposition following a particle
interaction results in a measurable temperature rise.
The temperature sensors developed for CRESST are superconducting
phase transition thermometers - transition edge sensors (TES) - consisting of thin tungsten films
evaporated onto a surface of the absorbers. The thermometers are stabilized in the
transition from the normal conducting to the superconducting phase where a
small temperature rise leads to a relatively large increase in resistance,
making them extremely sensitive thermometers.
In CRESST, absorber linked to such thermometers are operated as cryogenic calorimeter detectors that are very sensitive and can measure the total energy deposited by an
interacting particle.
|
|

Schematic drawing of a CRESST calorimeter element

A typical transition curve. Since it is very steep,
a small change in temperature results in a measurable change of resistance.
|
Detector Modules
In the first phase of the CRESST experiment (CRESST I), sapphire crystals were used
as target material. As explained above, energy deposited in these crystal via a particle interaction is visible as a heat signal read out with a
TES.
The detector modules for CRESST II and CRESST III exploit the fact that
most common backgrounds will produce a certain amount of light in a scintillating material,
while on the other hand the sought-for dark matter induced recoils
will produce little or no light.
Thus, detectors were developed based on
scintillating CaWO4 crystals as absorbers. In this crystal a particle
interaction produces mainly heat in the form of phonons, as for
sapphire. In addition a small amount of the energy deposited is
emitted as scintillation light. A light absorber (usually silicon on sapphire) together with a second thermometer acts as light detector which is also operated as cryogenic
calorimeter. Both detectors are mounted close to
each other forming a detector module and are enclosed in a scintillating and reflective housing for an efficient light collection.
As the amount of light produced differs for different kinds of particles, most common backgrounds (e.g. electron recoils, alpha events..) can be
eliminated. In this way a very efficient active background discrimination can be achieved using the so-called light yield defined as the fraction of light and heat energy as discrimination
parameter:
In this schematic drawing the horizontal bands resulting from different event types are shown in the ligh yield - energy plane.
Per definition the light yield is normalized to 1 for gamma radiation with an energy of 122keV. Therefore, the light yield of betas and gammas
(electron recoils) is about 1. For alpha particles the light yield is about 0.22 and for nuclear recoils (O, Ca, W)
it is even lower (0.1 - 0.02).
(Eur. Phys. J. C (2014) 74 , arXiv:1401.3332).
Due to finite detector resolution, these bands can be separated to some extent.
|