Wednesday, 30 October 2013

Fiat LUX

A new episode of the dark matter detection saga was just broadcast live from Sanford Lab.  LUX is a new direct detection experiment located in a South Dakota mine, not far from Mt Rushmore. Today they presented their results based on the first 3 months of data taking.

LUX is very similar to its direct competitor Xenon100. They both use a dual-phase xenon target, and  a combination of scintillation and ionization photons to detect collisions of dark matter particles with nuclei, estimate their recoil energies, and separate dark-matter-like events from gamma- and beta-ray backgrounds.  The active target region of LUX  is about 4 times larger than that of Xenon100, but the shorter time of data taking means the amount of data collected so far is similar.  However, one advantage of LUX is better collection of photons appearing inside the  detector, which allows them to lower the energy threshold for detection down to 3 keV nuclear recoils (compared to the 6.6 keVnr threshold in the latest Xenon100 analysis). Thanks to this feature, they already managed to beat the Xenon100 sensitivity significantly, especially for low mass (below 10 GeV) dark matter which produces few photons when it scatters on nuclei. This last aspect was what made today's announcement so interesting. Recently there has been several (partly contradictory) reports of possible detection of low mass dark matter, the most serious of which was the 3 events seen by the CDMS silicon detector. These signals were in tension with the  Xenon100 results, but the low mass region is experimentally difficult and an independent confirmation was more than welcome. If the CDMS signal was really dark matter LUX should be literally swamped with dark matter events. Instead, what they see is this:

There's 160 events surviving the experimental  cuts. They are plotted according to how many scintillation (S1) and ionization (S2) photons they produce. The background is expected to  show up in the blue band, and the dark matter signal (characterized by a lower ratio of ionization to scintillation) in the red band. What can be found in the red band is perfectly consistent with leakage from the background region (note in particular that there is no events below the center of the band), with  the p-value for the  background only hypothesis at the fairly large value of 35%. Translating that into limits on the scattering cross section of vanilla-type dark matter on nucleons results in this plot: 

The red line is the previous best limit from Xenon100. The blue line is the current 90% CL limit from LUX,  which puts them at the pole position in the entire mass range above GeV. They are the first to break the 10^-45 cm^2 cross section barrier: the limit goes down to 7.6*10^-46 cm^2  for dark matter mass of 33 GeV. To put it into perspective, the LHC can currently study processes with a cross section down to 10^-39 cm^2 (1 femtobarn).  The inlay shows the low mass region where positive signals were claimed by CDMS-Si (green), CoGeNT (orange), CRESST (yellow) and DAMA (grey). All of these regions are now comfortably excluded, at least in the context of simple models of dark matter.   

So,  the light dark matter signal that has been hanging around for several years is basically dead now.  Of course, theorists will try to reconcile the existing positive and negative results, just because it's their job.  For example, by playing with the relative couplings of dark matter to protons and neutrons one can cook up xenophobic models where dark matter couples much more strongly to silicon and germanium than to xenon.  But seriously, there's now little reason to believe that we are on the verge of a discovery. Next time, maybe.  

The LUX paper is here.