Search for the Neutron Decay $n\to X+\gamma$, Where $X$ is a Dark Matter Particle

Fornal and Grinstein recently proposed that the discrepancy between two different methods of neutron lifetime measurements, the beam and bottle methods, can be explained by a previously unobserved dark matter decay mode, $n\to X+\gamma$. We perform a search for this decay mode over the allowed range of energies of the monoenergetic $\gamma$ ray for $X$ to be dark matter. A Compton-suppressed high-purity germanium detector is used to identify $\gamma$ rays from neutron decay in a nickel-phosphorous-coated stainless-steel bottle. A combination of Monte Carlo and radioactive source calibrations is used to determine the absolute efficiency for detecting $\gamma$ rays arising from the dark matter decay mode. We exclude the possibility of a sufficiently strong branch to explain the lifetime discrepancy with 97% confidence.

Figure 1

The UCN bottle is installed in the existing beam line, and a HPGe detector is placed next to the outer wall of the vessel. The UCN gate valve is located upstream of the polarizing solenoid magnet. On the left is a photograph of our setup and on the right is a schematic.

Figure 2

Measured and simulated spectra in the ROI (white background). The blue and red lines show the Compton-scattering-suppressed spectra for the measurement with UCN and background measurement, respectively. The dotted line shows the simulated spectra from UCN capture and related $\gamma$ rays. The grey and thin black curves show the net UCN signal and the net signal after capture $\gamma$ subtraction, respectively. The peak plotted with a thick black line centered at 1200 keV shows an example of the size of the proposed dark matter (DM) decay that would be needed to explain the anomaly.

Figure 3

The data (solid circles) plotted are the sum of counts in a 12 keV ($\sim 3$ Gaussian peak widths) window above a piecewise fitted background across the ROI. The integration region is centered in a 100 keV wide fitting window, and it is performed in 4 keV steps. The unshaded center of the plot is the ROI. The predicted signal (solid line and shaded region) accounts for the energy dependent photopeak efficiency predicted by GEANT4 [16]. The open circles show the likelihood of the predicted signal in each overlapping bin.