Decay of krypton-85 could mask the results of the XENON100 detector.

Time is running out for Elena Aprile, a physicist at Columbia University in New York. In mid-March, she is scheduled to present the results of a dark-matter experiment at a workshop in Venice, Italy, but she doesn't have the results yet.

She and the rest of her team have blinded themselves to the data to avoid bias, while they wrestle with an analysis that needs to take account of radioactive contamination in their detector.

The XENON100 detector involves 161 kilograms of liquid xenon placed 1.4 kilometres underground at Gran Sasso National Laboratory near L'Aquila, Italy. It is designed to search for weakly interacting massive particles (WIMPs), a leading candidate for dark matter — which is believed to make up about one quarter of the Universe. The researchers believe that WIMPs pass through Earth in a constant stream and have a small chance of colliding with the nuclei of xenon atoms, producing charged particles and flashes of light that can be picked up by sensitive detectors.

Experiment's kryptonite

But there is a problem that threatens to make their results much harder to interpret. The commercially supplied xenon used in the experiment is contaminated with the radioactive isotope krypton-85, which comes from atmospheric nuclear tests and makes its way into the xenon which is produced by liquifying air and extracting xenon. The krypton isotope emits β- and γ-rays that could be mistaken for WIMPs. "This is one of the major challenges for large-scale xenon experiments," Aprile acknowledges.

The contamination of krypton in the commercially supplied xenon is at a concentration of a few parts per billion, and the team has used distillation — because xenon and krypton have different boiling points — to reduce it to around a hundred parts per trillion.

Their most recent run faced two obstacles. First, the level of concentration remains higher than hoped because they were unable to get the distillation column to match the level of purification claimed by a competitor experiment, XMASS in Japan, that is due to start taking data this year2.

Second, and more significantly, the team is having trouble estimating exactly how much krypton-85 is in the experiment. "The lower the concentration, the harder it becomes to measure," says Guillaume Plante, a graduate student at Columbia who works on the experiment. That means that, if they see a signal in the data, they could find themselves unable to tell whether it comes from a WIMP, or from krypton-85 radioactive decay.

In the last data release, an 11-day run in 2009 that did not see any WIMPs1, krypton-85 accounted for perhaps half of the 22 background events seen, says Antonio Melgarejo, a postdoc at Columbia who works on the experiment. Team members aren't yet certain how many background events were seen in the most recent run, of 100 days, but given the performance of the column it is likely to be hundreds.

The experiment's detectors can distinguish radioactive decays from WIMPs because they produce different ratio of charged particles to light, but only with an efficiency of about 99.5%, meaning that one out of every 200 of the background events could be interpreted as a false positive. That means the team could well end up in a similar situation to what happened in 2010 to the Cryogenic Dark Matter Search (CDMS II) experiment in the Soudan Underground Laboratory in Minnesota, which uses germanium instead of xenon for its detector. CDMSII detected two candidate WIMPs, but ultimately concluded that they were indistinguishable from background, turning a potentially exciting situation into a null result3.

The risk with high background levels is that, "you may reach a point where no one believes your ability to subtract off the background," says Sunil Golwala, a physicist at the California Institute of Technology in Pasadena who works on CDMS II.

Recruitment drive

Richard Gaitskell, the principal investigator of LUX, a xenon-based detector that is expected to record its first data towards the end of 2011 at the Sanford Underground Laboratory at the Homestake gold mine in South Dakota, says that it shouldn't be difficult to pick out a genuine signal. "If a dark matter experiment is working, the result should be trivial because the background signal is so low," he says. He estimates that the krypton concentration in LUX's xenon is less than ten parts per trillion, and that, even in a 300-day run, the experiment should not pick up any false positives due to krypton contamination.

Aprile has recruited additional team members to work on measuring the level of krypton. At the moment, the team's estimate is based on looking for one of krypton-85's characteristic decay signals -- a beta ray followed by a gamma ray -- but they would prefer to measure it directly by sampling the detector fluid. Atomic physicist Tanya Zelevinsky, one of Aprile's colleagues at Columbia, is working on an atom trap that could count krypton atoms one at a time, and so provide accuracy to the level of parts per trillion.

Henrique Araújo, the principal investigator of the ZEPLIN-III experiment at the Boulby Underground Lab, UK, which also uses xenon, says the uncertainty underlines the need to have several dark-matter experiments running in parallel. ZEPLIN-III uses less xenon than XENON100, meaning it is less likely to see WIMPs, but it is better at distinguishing radioactive background from a potential dark matter signal, he says. "This is a very competitive area of experimental physics and everyone wants to be the first to find it," he adds.

The project's latest paper is largely written, says Aprile, and she's still hoping to be ready in time for the Venice meeting. "All we're waiting for are the data," she insists.