
3 CCDS TO SEARCH FOR DM
2 Direct detection signals from low-mass DM particles
For DM with mχ < 1 GeV c−2, the traditional approach to search for the energy deposited in the
target by nuclear recoils from DM scattering becomes limited because the mismatch between the
mass of the DM particle and the nucleus does not allow for efficient transfer of energy. Recently,
other direct-detection mechanisms have been proposed to extend sensitivity to lower DM masses.
The first is to search for electrons or photons emitted in the DM-nucleus interaction, which can
carry a significant fraction of the DM particle’s kinetic energy and appear above the detection
threshold. The “Migdal effect” [6]is the process where an electron from the recoiling atom is
emitted, with a probability in silicon of O(10−5)[7]. The probability of photon emission (the
“Bremsstrahlung” process) is significantly smaller [8]. Another possibility is to look directly for
the scattering of low-mass DM particles with electrons in the target [9,10]. Atomic electrons
are lighter than nuclei and, since they have a momentum distribution, there are regions of phase
space where the electron can take a significant fraction of the DM particle’s kinetic energy. DM-
e interactions naturally arise in hidden-sector vector-portal DM theories [11], where the DM-e
scattering is mediated by a “hidden photon.” The hidden photon itself could also constitute the
DM and be absorbed by atomic electrons [12–14]. Since all these DM signals are electronic recoils
in the target, several experiments have placed exclusion limits on sub-GeV DM from their measured
electronic-recoil spectra [15–18].
The kinetic energy of a DM particle in the galactic halo is EK∼10−6mχc2. Most often, only a
fraction of EKis transferred to the electron, with the predicted deposited-energy spectrum highly
peaked at the lowest energies. Fig. 1shows the spectrum from a 1 GeV c−2DM particle scattering
with electrons in silicon predicted by the EXCEED-DM code [19]. Most of the spectrum—even at
1 GeV c−2—is below 10 eVee, which corresponds to only a few ionized charges in silicon. Thus,
detectors capable of counting single ionized electrons (already achieved in semiconductor and
noble-element direct-detection technologies) are needed to perform sensitive searches for low-
mass DM-e interactions. Semiconductor detectors have a fundamental advantage since the ion-
ization threshold is ∼1 eV compared to >10 eV for noble elements. Furthermore, the momentum
distribution of the electrons in the semiconductor valence band are better “kinematically matched”
to the DM flux [10]. For this reason, the current sensitivity of silicon CCD detectors to DM-e scat-
tering surpasses that of noble liquids for most of the parameter space (specially for smaller DM
masses and lighter mediators) even though their target exposures are orders of magnitude smaller
(O(g·day) vs. O(tonne·day)).
3 CCDs to search for DM
Charge-coupled devices (CCDs) are monolithic solid-state silicon imaging devices that measure
the free charges generated in their fully-depleted active target. Particles generate free charges (e-
h pairs) in the CCD active region by ionization, with a minimum energy to ionize an e-h pair of 1.2
eV and one e-h pair generated on average for every 3.8 eV of kinetic energy deposited by a recoiling
electron. The free charges are then drifted by an electric field toward the pixel array. Since charge
diffuses laterally with time as it drifts, the spread of the charge cluster in the image (x-yplane) is
positively correlated to the depth (z) of the interaction. Thus, CCDs can, in principle, provide the
deposited energy and (x,y,z)location of particle interactions in the bulk silicon. In reality, the
spatial resolution may be limited in one or more axes depending on the readout mode and event
2