Inelastic Dark Matter in Light of DAMA/LIBRA

TL;DR

The paper addresses reconciling the DAMA/LIBRA annual modulation with null results from multiple direct-detection experiments by proposing inelastic dark matter (iDM) with a small excited-state splitting δ ~ 100 keV. It develops the theoretical framework, kinematic consequences, and rate calculations for iDM, demonstrating that heavier targets, enhanced modulation, and suppression of low-energy events can align DAMA with other experiments. Through detailed analysis of experimental limits (DAMA, CDMS, XENON, ZEPLIN, KIMS, CRESST) and benchmark points, it identifies viable parameter space (mχ ~ 50–200 GeV, δ ~ 120–134 keV) driven largely by CRESST tungsten data, and makes falsifiable predictions for future searches, especially in the 20–60 keV recoil range and tungsten channels. The work emphasizes the importance of high-energy and modulation-sensitive analyses and outlines concrete experimental directions to confirm or exclude the iDM scenario.

Abstract

Inelastic dark matter, in which WIMP-nucleus scatterings occur through a transition to an excited WIMP state ~ 100 keV above the ground state, provides a compelling explanation of the DAMA annual modulation signal. We demonstrate that the relative sensitivities of various dark matter direct detection experiments are modified such that the DAMA annual modulation signal can be reconciled with the absence of a reported signal at CDMS-Soudan, XENON10, ZEPLIN, CRESST, and KIMS for inelastic WIMPs with masses O(100 GeV). We review the status of these experiments, and make predictions for upcoming ones. In particular, we note that inelastic dark matter leads to highly suppressed signals at low energy, with most events typically occurring between 20 to 45 keV (unquenched) at xenon and iodine experiments, and generally no events at low (~ 10 keV) energies. Suppressing the background in this high energy region is essential to testing this scenario. The recent CRESST data suggest seven observed tungsten events, which is consistent with expectations from this model. If the tungsten signal persists at future CRESST runs, it would provide compelling evidence for inelastic dark matter, while its absence should exclude it.

Figures (10)

• Figure 1: The spectra of modulated events at DAMA for a 100 GeV WIMP with $\delta = 120$ keV. The unmodulated spectrum is similar.
• Figure 2: The modulated fraction of events as a function of $\delta$ (in keV) for $m_\chi = 100$ GeV.
• Figure 3: The contours to the right of which no particles are present in the halo which can scatter at CDMS for $v_{esc}=500 \ {\rm km/s}$ (solid) and $v_{esc}= 600 \ {\rm km/s}$ (dashed), in the $\delta -m_\chi$ plane.
• Figure 4: Modulation amplitudes for benchmark points at DAMA for $v_{esc}=500\; {\rm km/s}$. The measured modulation amplitudes of the full DAMA + DAMA/LIBRA data set are shown as data poins. The dashed line is the 70 GeV benchmark, which is outside the 90% confidence of the combined DAMA + DAMA/LIBRA 2-6 keV modulation, but inside the 90% confidence region for DAMA/LIBRA 2-6 keV rates alone. For the solid lines, from the highest to lowest peak, and narrowest to broadest curves, $m_\chi = 90,120,150,180,250 \;{\rm GeV}$.
• Figure 5: Spectrum of events at CDMS (Ge) for $v_{esc}=500\; {\rm km/s}$ (from lowest to highest signal) $m_\chi = 120,150,180,250 \; {\rm GeV}$. Curves for the other benchmark points are zero.