Semiconductor Probes of Light Dark Matter

TL;DR

This work shows that semiconductor detectors with single-electron sensitivity can probe light dark matter through DM–electron interactions, focusing on electric and magnetic dipole moments as well as broken U(1) gauge scenarios. It derives detection rates for both isolated atoms and semiconductor valence-band electrons, incorporating Coulomb enhancement and lattice effects, and translates these rates into 95% CL sensitivities for CDMSLite using Ge, Si, and Xe targets. The analysis demonstrates that semiconductors substantially outperform noble gases at low recoil energies, enabling exploration of dipole moment scales up to roughly $10^3$ TeV for DM masses in the 0.6 MeV–10 GeV range, and highlights the potential reach of light mediators in U(1) models. Overall, the paper argues that sub-GeV DM with dipole or new gauge interactions is a well-motivated target and that CDMSLite-like semiconductor experiments can access previously unexplored parameter space with significant practical impact.

Abstract

Dark matter with mass below about a GeV is essentially unobservable in conventional direct detection experiments. However, newly proposed technology will allow the detection of single electron events in semiconductor materials with significantly lowered thresholds. This would allow detection of dark matter as light as an MeV in mass. Compared to other detection technologies, semiconductors allow enhanced sensitivity because of their low ionization energy around an eV. Such detectors would be particularly sensitive to dark matter with electric and magnetic dipole moments, with a reach many orders of magnitude beyond current bounds. Observable dipole moment interactions can be generated by new particles with masses as great as 1000 TeV, providing a window to scales beyond the reach of current colliders.

Figures (6)

• Figure 1: One-loop contributions to DM dipole moment due to a charged fermion-scalar pair.
• Figure 2: One-loop contribution to DM mass due to a charged fermion-scalar pair.
• Figure 3: Density of states as a function of binding energy for the valence bands of germanium (red) and silicon (blue), normalized such that $\int dE_B \, \rho(E_B)=1$.
• Figure 4: Exclusion sensitivity at 95% confidence level possible after 1 year, for (a) electric and (b) magnetic dipole moments. The solid lines assume a background of 1 event/day/kg/keV, while the dashed lines assume no background. Areas above the curves for germanium (red), silicon (blue), and xenon (brown) would be excluded. Regions in gray are already excluded for all models of DM by other experiments or astrophysical data. Masses to the left of the dashed black line are potentially constrained by supernova cooling and BBN. While a detailed calculation of these constraints on lighter masses is beyond the scope of this work, it is unlikely the entire region is fully excluded.
• Figure 5: Exclusion sensitivity at 95% confidence level possible after 1 year, for effective $U(1)$ coupling $\left(\lambda = \epsilon \sqrt{\frac{g_\chi^2}{4\pi}} \right)$ with (a) $m_A = 10$ MeV and (b) $m_A = 1$ meV. The solid lines assume a background of 1 event/day/kg/keV, while the dashed lines assume no background. Areas above the curves for germanium (red), silicon (blue), and xenon (brown) would be excluded. Regions in gray are already excluded for all models of DM by other experiments or astrophysical data. Masses to the left of the dashed black line are potentially constrained by supernova cooling and BBN. While a detailed calculation of these constraints on lighter masses is beyond the scope of this work, it is unlikely the entire region is fully excluded.