The longest standing potential signal in this region was first reported by the DAMA/NaI collaboration more than 10 years ago [196]. The current iteration of this experiment, DAMA/LIBRA, consists of 250 kg of extremely radiopure thallium-doped NaI scintilla- tor crystals. DAMA/LIBRA has the largest exposure of any dark matter data, with nearly 1.2 ton yrs of data collected over 13 annual cycles [187]. DAMA measures only the scin- tillation light produced by each particle interaction, so it cannot distinguish between the expected nuclear-recoil signal from WIMPs and electron-recoil backgrounds on an event- by-event basis. Instead, it attempts to identify WIMP interactions through the presence of a residual annually modulating signal after removing the unmodulated rate. This task is made easier due to the high radiopurity of DAMA’s detectors, leading to a low rate of un- modulated backgrounds in the low energy region (.1 event keV−1ee kg−1 day−1). DAMA’s energy threshold of 2 keVeeand target containing a light nucleus (Na) also make it sensitive to recoils from very low mass WIMPs, which produce recoil energies that may fall below the threshold of other experiments. DAMA’s energy scale is given in keVee units above, denoting “electron-equivalent” energy. Since nuclear recoils produce less scintillation light than electrons for the same total recoil energy, an electron-equivalent scale is typically used, where the recoil energy (in units of keVnr is given byE/q forE measured in keVee. Here, q is a target-dependent “scintillation yield” for nuclear recoils, for which measurements by the DAMA collaboration indicate qN a= 0.30±0.01 andqI = 0.09±0.01 [197].
As shown in Fig. 2.4, measurements of the residual counting rate in DAMA/LIBRA indicate the presence of an annual modulation at 8.9σ significance. This modulation has a phase consistent with that expected for a dark matter signal (t0 = 144±8 days [187], while t0 = 152.5 days in the SHM). As shown in Fig. 2.4b, the modulation peaks in the lowest energy bins and is absent at higher energies. It also occurs only for events in- teracting in a single detector module and is absent for multiple-detector hits. Although many suggestions have been proposed for backgrounds which could produce these features (e.g., [198–200]), to date the signal reported by DAMA/LIBRA remains unexplained by such a background [201] and has been interpreted as evidence for WIMP interactions. However, for spin-independent elastic scattering of WIMPs under standard assumptions, experiments such as XENON100 [54] and CDMS II [153] exclude the parameter space favored by the
Energy (keV) Sm (cpd/kg/keV)
-0.05 -0.025 0 0.025 0.05
0 2 4 6 8 10 12 14 16 18 20 2-6 keV
Time (day)
Residuals (cpd/kg/keV)
DAMA/NaI (0.29 ton yr)
(target mass = 87.3 kg) DAMA/LIBRA (0.53 ton yr) (target mass = 232.8 kg)
a)
b)
Figure 2.4: a) Residual counting rate for single-detector hits versus time for DAMA/NaI and DAMA/LIBRA from 2–6 keVee, after subtracting the mean total counting rate in each year. The horizontal axis gives the number of days since January 1st of the first year that DAMA/NaI operated. Clear evidence is seen for an annual modulation peaking in late May, over more than 10 annual cycles. b) Energy spectrum for the amplitude of the modulated rate. The modulation peaks in the lowest energy bins as expected for a WIMP signal, with no evidence for modulation above 8 keVee. Figure from Bernabei et al. [188]
DAMA/LIBRA modulation by several orders of magnitude if the measured recoil rate is primarily from I, as would be expected for WIMPs with masses &20 GeV. Although the low-mass region corresponding to Na scattering (where the much higher rate of I recoils is missed since it falls below threshold) is also currently disfavored [144, 156], the exclusion in this region is less stringent, and uncertainties in the halo model or detector response at low energy may allow compatibility [82, 194, 195].
2.1.4.2 CoGeNT
The CoGeNT collaboration has developed a second detector technology with excellent sen- sitivity to WIMPs with masses .10 GeV, but with a target and measurement technique different than that employed by DAMA, in order to provide a cross check on their possi- ble low-mass WIMP signal. CoGeNT operates a 440 g P-type point contact germanium ionization detector in the Soudan Underground Laboratory [189, 193]. Due to its point con- tact electrode, the capacitance of the detector is significantly reduced relative to standard
a) b)
Figure 2.5: a) Total counting rate in CoGeNT at low energy after applying the surface event rejection criteria. A residual exponential excess remains after subtracting the constant Compton scatter background and known L-shell activation peaks near 1.3 keVee. This excess is similar to the spectrum expected for a WIMP with mχ ∼ 5–10 GeV and σSI ∼ 10−41–10−40cm2, as shown in the inset. b) Total counting rate versus time in CoGeNT after removing surface events and subtracting the Compton and L-shell backgrounds. Evidence for annual modulation with a phase consistent with that expected for a dark matter signal is observed, with a significance of∼2.8σ in the full energy range, limited by statistics. Figure from Aalseth et al. [189]
cylindrical geometries, allowing an extremely low ionization energy threshold of 0.4 keVee
in a nearly half-kg detector (forqGe∼0.2 at these energies, this corresponds to a∼2 keVnr recoil-energy threshold). This geometry also allows the identification of interactions occur- ring within ∼1 mm of the detector surface to be identified due to the slower rising pulses for such events. This gives the ability to reject external low-energy backgrounds, which typically have a penetration depth1 mm in the energy range of interest.
After applying their surface event rejection cut, CoGeNT initially reported a large ex- ponential excess of events from 0.5–1 keVee that could not be accounted for by known backgrounds [193], but which could be explained by a WIMP with a mass and cross sec- tion roughly compatible with that needed to explain the DAMA/LIBRA results described above [82]. Further data taking confirmed this excess and gave weak evidence (∼2.8σ, lim- ited by statistics) for a ∼15% annual modulation in the counting rate in this excess [189], with a phase and spectrum consistent with that found by DAMA/LIBRA [190, 202]. This excess and evidence for a corresponding modulation is shown in Fig. 2.5. However, it has re-
cently been suggested that a significant fraction (∼75%) of the exponential excess observed by CoGeNT at low energy is due to leakage of surface electron recoils that cannot be com- pletely distinguished from bulk events at low energy [190–192]. In this case, the parameter space consistent with a WIMP signal providing the remaining∼25% of the excess is pushed to WIMP masses,mχ≈10 GeV andσSI ≈2×10−41, improving agreement with the excess reported by CRESST-II (described in Sec. 2.1.4.3 below), but decreasing agreement with DAMA/LIBRA unless ion-channeling or local halo substructure is significant [176, 190].
If the modulation reported by CoGeNT is also due to WIMPs, then a modulation fraction of the remaining total WIMP rate of order 50–100% would be required, which is an order of magnitude larger than expected in the SHM. This modulation also appears to extend to higher energies than would be compatible with the measured total rate [202, 203], suggesting either a background origin or statistical fluctuation for the modulation detected above∼1.2 keVee. This is supported by a search for a corresponding modulation from 1.2–3 keVee in CDMS, which finds no significant evidence for modulation [203] at high energy, although the CDMS modulation analysis does not constrain the region below 1.2 keVee where the WIMP signal would be expected. CoGeNT is continuing to acquire statistics with its existing detector module and plans an upgrade of its detector mass by roughly an order of magnitude. This larger data set will give an improved measurement of the modulation and possibly identify or exclude a background origin for the residual low-energy excess.