The goal of CDMS is to detect dark matter, which has only been detected through large-scale gravitational interactions.  Direct detection is yet to occur in an Earth-bound laboratory.  The dark matter particles are thought to be between ten and one thousand GeV.  CDMS uses cryogenic germanium and silicon detectors, which are capable of detecting weakly interactive dark matter (WIMPs). 

WIMPs are detected through their interactions with the nuclei in the germanium.  When a nucleus is hit, it recoils, causing the whole germanium crystal to vibrate.  These vibrations, or phonons, propagate to the surface of the crystal where they heat sensors consisting of thin aluminum traps connected by tungsten meanders.  The tungsten is held at critical temperature.  When the phonons reach the aluminum, they excite quasi-particle states, which propagate to the tungsten and heat it up. 

When the quasi-particle states reach the tungsten, the temperature of the tungsten rises, and therefore so does the resistance of the circuit.  This causes the bias current to decrease since the voltage across the tungsten is held constant.   The resulting pulse is picked up by SQUID (Superconducting QUantum Interference Device) amplifiers. 

There are other particles that will go through the detectors besides WIMPs.  The CDMS detectors are shielded to minimize the number of these other particles.  The detectors are capable of discriminating between most of them and the WIMP signal.

The detectors will run in the Soudan Mine in Minnesota, about a half-mile underground.  The deep site is chosen to shield the detectors from cosmic rays.  Nevertheless, electrons and photons from the rock and the shield will still be present underground.  When these particles hit the detectors, they most often recoil off of the electrons in the germanium and produce a large amount of ionization.  This process creates more ionization per unit of phonon energy than a recoiling of the nucleus from a WIMP interaction.  So if we examined a graph of the ionization produced vs. phonon energy produced, we would see two distinct bands.  Those particles in the top band are electrons or photons or something else – not WIMPs – because they have relatively high ionization.  Those in the lower band, where there is less ionization, are candidates for WIMPs.

There is another possibility for the identity of the particles in the lower band, however.  They could be neutrons.  Like WIMPs, they interact with the germanium nucleus, which results in a nuclear recoil.  Therefore they would produce a signal indistinguishable from that of a WIMP.  But there are ways of discriminating between WIMPs and neutrons, just as there are ways to distinguish between WIMPs from electrons. 

Separating the neutron signal is the reason CDMS uses silicon detectors.  Silicon has an atomic number of 28, compared to germanium’s much larger 73.  Due to the larger germanium nucleus, WIMPs are much more likely to interact in the germanium than they are in the silicon (about seven times as likely).  This is because their interactions are governed by the weak force and the cross-section, or probability of interaction depends strongly on the atomic number.  Neutrons, on the other hand, interact through the strong force, which does not depend so strongly on the size of the nucleus.  Consequently, neutrons are as likely to interact in a silicon detector, as they are a germanium nucleus of the same volume.

This property of the silicon detectors allows us to use them to measure the flux of neutrons, and hence, the number of background neutron events.   Subtracting these events from the total signal allows us to determine whether there is a WIMP signal and to determine how large it is.  If there is no WIMP signal, we should see an equal number of pulses on both silicon and germanium detectors.  If there are more pulses in the germanium detectors than in the silicon, then the extra events would correspond to the WIMP signal.  In other words, Germanium Events – Silicon Events = WIMP Events.

There is also another method for sorting out neutrons since CDMS will employ many detectors of each type.    A certain percentage of neutrons should interact with more than one detector while WIMPs will hit one detector and only one.  We can measure the neutron flux by counting how many double scatter events there occur.  This number can then be used in the same way that the flux of neutrons determined from the silicon detectors is used to extract the strength of the WIMP signal.