Constraining Light Dark Matter with Low-Energy e+e- Colliders

Abstract

We investigate the power of low-energy, high-luminosity electron--positron colliders to probe hidden sectors with a mass below ~10 GeV that couple to Standard Model particles through a light mediator. Such sectors provide well-motivated dark matter candidates, and can give rise to distinctive mono-photon signals at B-factories and similar experiments. We use data from an existing mono-photon search by BaBar to place new constraints on this class of models, and give projections for the sensitivity of a similar search at a future B-factory such as Belle II. We find that the sensitivity of such searches are more powerful than searches at other collider or fixed-target facilities for hidden-sector mediators and particles with masses between a few hundred MeV and 10 GeV. Mediators produced on-shell and decaying invisibly to hidden-sector particles such as dark matter can be probed particularly well. Sensitivity to light dark matter produced through an off-shell mediator is more limited, but may be improved with a better control of backgrounds, allowing background estimation and a search for kinematic edges. We compare our results to existing and future direct detection experiments and show that low-energy colliders provide an indispensable and complementary avenue to search for light dark matter. The implementation of a mono-photon trigger at Belle II would provide an unparalleled window into such light hidden sectors.

Figures (10)

• Figure 1: $\gamma + \slashed E$ production channels for LDM coupled through a light mediator. Left: Resonant $\Upsilon(3S)$ production, followed by decay to $\gamma + \chi \, \overline \chi$ through an on- or off-shell mediator. Right: The focus of this paper -- non-resonant $\gamma + \chi \, \overline \chi$ production in $e^+ e^-$ collisions, through an on- or off-shell light mediator $A'^{(*)}$. (Note that in this paper, the symbol $A'$ is used for vector, pseudo-vector, scalar, and pseudo-scalar mediators.)
• Figure 2: Regions in the $m_{\chi}$--$m_{A'}$ plane with different characteristic $\gamma$+$\slashed E$ signals ($A'$ is any type of mediator). Region (a) corresponds to an off-shell heavy mediator, for which an effective operator analysis holds if $m_{A'} \gg \sqrt{s}$. In Region (b) the mediator is invisible and is produced on-shell. In Region (c), while the mediator is light enough to be produced on shell, $\chi \overline \chi$ production occurs through an off-shell mediator.
• Figure 3: Typical simulations of $\gamma$+$\slashed E$ signals compared to data that was scanned from the B A B A R Collaboration (unpublished) Aubert:2008as, in both the "High-E" ( left) and "Low-E" ( right) search regions (where $3.2 \text{ GeV}<E_\gamma^*<5.5 \text{ GeV}$ and $2.2 \text{ GeV}<E_\gamma^*<3.7 \text{ GeV}$, respectively; see Sec. \ref{['sec:BaBar']} for more details). The red histogram illustrates $\chi \overline \chi$ production through an off-shell heavy mediator (region (a)), resulting in a rising spectrum. The histogram corresponds to $m_\chi=1 \text{ GeV}$ and $m_{A'}=12 \text{ GeV}$. The orange histograms show the peaked spectra arising from on-shell production of an invisible mediator (region (b)), with $m_{A'}=0.5 \text{ MeV}$ (left) or $4 \text{ GeV}$ (right). The green histogram shows the typical broad spectrum resulting from $\chi \overline \chi$ production through an off-shell light mediator (region (c)) (we show $m_\chi=m_{A'}=1 \text{ GeV}$). In each case the cross-section is scaled to lie at the 95% CL limits presented in Sec. \ref{['sec:BaBar']}.
• Figure 4: Lower bounds on $m_{A'}/\sqrt{g_e g_\chi}$ in region (a) of Fig. \ref{['fig:search-regions']} (production of $\chi \overline \chi$ through a heavy off-shell mediator), for ( left) a fixed DM mass of 10 MeV, and ( right) a fixed mediator mass of 12 GeV. The solid black line / blue shaded region show the bounds from B A B A R data (this work) with a vector mediator. On the right, the bounds with other mediators are shown with different line styles, while on the left they are almost identical to the vector case and thus not shown separately. The solid and dotted blue line both show the projected reach of Belle II in the vector-mediated case assuming that the various background components are known at the $5-20\%$ level ("systematics" limited) or, more idealistically, is known perfectly up to statistical fluctuations ("statistics" limited) (see Sec. \ref{['sec:Belle-II-projections']} for details). The gray shaded region is excluded by combining LEP bounds Fox:2011fx with $g_\chi$-perturbativity. For the hidden photon case, this limit is strengthened by including $Z$-pole constraints Hook:2010tw on $\varepsilon$, as shown by the green line. See text for more details.
• Figure 5: Upper bounds on the coupling of electrons to a mediator decaying invisibly to dark-sector states (region (b) of Fig. \ref{['fig:search-regions']}). The solid black line / blue shaded region shows the bound from B A B A R data (this work), for a vector or pseudo-vector mediator. The dotted line shows the bound for a scalar or pseudo-scalar mediator. The black dashed line shows the projected upper limit from an "improved B A B A R" analysis for a vector or pseudo-vector mediator, where the $\gamma\slashed\gamma$ background has been reduced by a factor of 10. The projected reaches of four possible searches for a vector mediator at Belle II are shown by the solid blue lines: a converted mono-photon search (dashed, labelled (a) and (b), which respectively assume no (a factor of 10) improvement in the $\gamma\slashed\gamma$ background rejection over the "improved B A B A R" projection), a standard mono-photon search (solid), and a low-energy mono-photon search (dot-dashed) (see Sec. \ref{['sec:Belle-II-projections']}). The gray shaded region is excluded by LEP Fox:2011fx. Additional limits relevant for sub-GeV mediators are shown in Fig. \ref{["fig:A'-invisible"]}. See text for more details.