Analysis of the low-energy electron-recoil spectrum of the CDMS experiment

We report on the analysis of the low-energy electron-recoil spectrum from the CDMS II experiment using data with an exposure of 443.2 kg-days. The analysis provides details on the observed counting rate and possible background sources in the energy range of 2–8.5 keV. We find no significant excess of a peaked contribution to the total counting rate above the background model, and compare this observation to the recent DAMA results. In the framework of a conversion of a dark matter particle into electromagnetic energy, our 90% confidence level upper limit of $0.246\mathrm{\text{events}}/\mathrm{kg}/\mathrm{day}$ at 3.15 keV is lower than the total rate above background observed by DAMA. In absence of any specific particle physics model to provide the scaling in cross section between NaI and Ge, we assume a $Z^2$ scaling. With this assumption the observed rate in DAMA remains higher than the upper limit in CDMS. Under the conservative assumption that the modulation amplitude is 6% of the total rate we obtain upper limits on the modulation amplitude a factor of $\sim 2$ lower than observed by DAMA, constraining some possible interpretations of this modulation.

Figure 1

Fit (red line) to the efficiency corrected low-energy spectrum consisting of a background model (gray/dashed) and three Gaussian distribution functions describing the 10.36 keV line from ${}^{71}\mathrm{Ge}$ (black), the 8.98 keV line from ${}^{65}\mathrm{Zn}$ (blue) and a line at the energy of ${}^{55}\mathrm{Mn}$ (green, see text). The total counting rate of the latter two lines is given in the figure.

Figure 2

Fit (red) to the published DAMA/LIBRA low-energy spectrum [3], consisting of a background model (gray/dashed) and a Gaussian distribution function (green). The parameters of the Gaussian are given in the figure.

Figure 3

90% CL upper limit on the total counting rate in Ge from this analysis (black/solid). The corresponding upper limit on the total counting rate in NaI under the assumption of a $Z^2$ scaling of the conversion cross section (see text) is also shown (blue/dashed). The black data point with $2\sigma$ error bars gives the total counting rate of the 3.15 keV peak of DAMA/LIBRA derived from a fit to their spectrum (see Fig. 2). The insert compares the upper limit on the modulation amplitude assumed to be 6% of the unscaled upper limit (black/solid) and the $Z^2$ scaled upper limit in NaI (blue/dashed) with the $2\sigma$ regions of the annual modulation amplitude observed by DAMA ($\mathrm{NaI}+\mathrm{LIBRA}$) in the 2–4 keV (red/filled) and 2–6 keV (green/hatched) energy range.