Dark Matter Week: Do you feel lucky today?

[tweetmeme only_single=false service=”wp.me” source=”allinthegutter”]

First things first, we owe you an apology here at we are all in the gutter. Well, Stuart and I do, for not having followed up with our posts Wednesday and Thursday. We promised to tell you a little bit more about dark matter before the announcement today and alas, we sold you short. We still intend to do so, but in the meantime Friday caught up with us and the Cryogenic Dark Matter Search (CDMS) experiment have announced their results right on the mark and we felt we should tell you what they are.

They’ve published a really good summary here , which I’ll now take the liberty to quote because it’s rather clear.

First, a little bit about the experiment itself:

The Cryogenic Dark Matter Search (CDMS) experiment, located a half-mile underground at the Soudan mine in northern Minnesota, uses 30 detectors made of germanium and silicon in an attempt to detect such WIMP scatters. The detectors are cooled to temperatures very near absolute zero. Particle interactions in the crystalline detectors deposit energy in the form of heat,
and in the form of charges that move in an applied electric field. Special sensors detect these signals, which are then amplified and recorded in computers for later study. A comparison of the size and relative timing of these two signals can allow the experimenters to distinguish whether the particle that interacted in the crystal was a WIMP or one of the numerous known particles that come from radioactive decays, or from space in the form of cosmic rays. These background particles must be highly suppressed if we are to see a WIMP signal. Layers of shielding materials, as well as the half-mile of rock above the experiment, are used to provide such suppression.

The CDMS experiment has been searching for dark matter at Soudan since 2003. Previous data have not yielded evidence for WIMPs, but have provided assurance that the backgrounds have been suppressed to the level where as few as 1 WIMP interaction per year could have been detected.

We are now reporting on a new data set taken in 2007- 2008, which approximately doubles the sum of all past data sets.

One of the hardest things about such an experiment, is to ensure that you know your kit well enough that you can tell the real signal from background noise that come from other sources (i.e., not dark matter):

With each new data set, we must carefully evaluate the performance of each of the detectors, excluding periods when they were not operating properly. Detector operation is assessed by frequent exposure to sources of two types of radiation: gamma rays and neutrons. Gamma rays are the principal source of normal matter background in the experiment. Neutrons are the only type of normal matter particles that will interact with germanium nuclei in the billiard ball style that WIMPs would, although neutrons frequently scatter in more than one of our detectors. This calibration data is carefully studied to see how well a WIMP-like signal (produced by neutrons) can be seen over a background (produced by gamma rays). The expectation is that no more than 1 background event would be expected to be visible in the region of the data where WIMPs should appear. Since background and signal regions overlap somewhat, achievement of this background level required us to throw out roughly 2/3 of the data that might contain WIMPs, because these data would contain too many background events.

A particularly interesting aspect of the data analysis, commonly used for this type of experiment, is that it is blind. The CDMS team explains:

All of the data analysis is done without looking at the data region that might contain WIMP events. This standard scientific technique, sometimes referred to as ‘blinding’, is used to avoid the unintentional bias that might lead one to keep events having some of the characteristics of WIMP interactions but that are really from background sources. After all of the data selection criteria have been completed, and detailed estimates of background ‘leakage’ into the WIMP signal region are made, we ‘open the box’ and see if there are any WIMP events present.

And so, what did they find? In short, they found two events that fit the bill. If these are indeed real events, this has been the first direct detection of Dark Matter in our scientific history. But alas, it’s never that simple. There’s a non-negligible chance, of around 23%, that these two events were created by background sources. I.e., that the signal hitting the detectors was a result of an interaction of particles which have nothing to do with dark matter. Of course, this means there’s a 77% chance that these two events were real, but 77% is generally not considered high enough to make a detection statistical significant in scientific terms. This is what the CDMS team explains:

In this new data set there are indeed 2 events seen with characteristics consistent with those expected from WIMPs. However, there is also a chance that both events could be due to background particles. Scientists have a strict set of criteria for determining whether a new discovery has been made, in essence that the ratio of signal to background events must be large enough that there is no reasonable doubt. Typically there must be less than one chance in a thousand of the signal being due to background. In this case, a signal of about 5 events would have met those criteria. We estimate that there is about a one in four chance to have seen two backgrounds events, so we can make no claim to have discovered WIMPs. Instead we say that the rate of WIMP interactions with nuclei must be less than a particular value that depends on the mass of the WIMP. The numerical values obtained for these interaction rates from this data set are more stringent than those obtained from previous data for most WIMP masses predicted by theories. Such upper limits are still quite valuable in eliminating a number of theories that might explain dark matter.

So why is everyone so excited, if the significance is not enough to break out the champagne? Simply put, because even though none of us is going to take this as scientific proof of dark matter detection, most of us also knows that real detections are often preceded by marginal detections. And that’s exciting. There’s a feeling that we’re really not that far, and 77% can feel very encouraging. In the words of my office mate – 77% has never been more exciting.