Wednesday 19 November 2014

Update on the bananas

One of the most interesting physics stories of this year was the discovery of an unidentified 3.5 keV x-ray  emission line from galactic clusters. This so-called bulbulon can be interpreted as a signal of a sterile neutrino dark matter particle decaying into an active neutrino and  a photon. Some time ago I wrote about the banana paper that questioned the dark matter origin of the signal. Much has happened since, and I owe you an update. The current experimental situation is summarized in this plot:

To be more specific, here's what's happening.

  •  Several groups searching for the 3.5 keV emission have reported negative results. One of those searched for the signal in dwarf galaxies, which offer a  much cleaner environment allowing for a more reliable detection. No signal was found, although the limits do not exclude conclusively the original bulbulon claim. Another study looked for the signal in multiple galaxies. Again, no signal was found, but this time the reported limits are in severe tension with the sterile neutrino interpretation of the bulbulon. Yet another study failed to find the 3.5 keV line in  Coma, Virgo and Ophiuchus clusters, although they detect it in the Perseus cluster. Finally, the banana group analyzed the morphology of the 3.5 keV emission from the Galactic center and Perseus and found it incompatible with dark matter decay.
  • The discussion about the existence of the 3.5 keV emission from the Andromeda galaxy is  ongoing. The conclusions seem to depend on the strategy to determine the continuum x-ray emission. Using data from the XMM satellite, the banana group fits the background in the 3-4 keV range  and does not find the line, whereas this paper argues it is more kosher to fit in the 2-8 keV range, in which case the line can be detected in exactly the same dataset. It is not obvious who is right, although the fact that the significance of the signal depends so strongly on the background fitting procedure is not encouraging. 
  • The main battle rages on around K-XVIII (X-n stands for the X atom stripped of n-1 electrons; thus, K-XVIII is the potassium ion with 2 electrons). This little bastard has emission lines at 3.47 keV and 3.51 keV which could account for the bulbulon signal. In the original paper, the bulbuline group invokes a model of plasma emission that allows them to constrain  the flux due to the K-XVIII emission from  the  measured ratios of the strong S-XVI/S-XV and Ca-XX/Ca-XIX lines. The banana paper argued that the bulbuline model is unrealistic as it  gives inconsistent predictions for some plasma line ratios. The bulbuline group pointed out that the banana group used wrong numbers to estimate the line emission strenghts. The banana group maintains that their conclusions still hold when the error is corrected. It all boils down to the question whether the allowed range for the K-XVIII emission strength assumed by the bulbine group is conservative enough. Explaining the 3.5 keV feature solely by K-XVIII requires assuming element abundance ratios that are very different than the solar one, which may or may not be realistic.   
  •  On the other hand, both groups have converged on the subject of chlorine. In the banana  paper it  was pointed out that the 3.5 keV line may be due to the Cl-XVII (hydrogen-like chlorine ion) Lyman-β transition which happens to be at 3.51 keV. However the bulbuline group subsequently derived limits on the corresponding Lyman-α line at 2.96 keV. From these limits, one can deduce in a fairly model-independent way that the contribution of Cl-XVII Lyman-β transition is negligible.   

To clarify the situation we need more replies to comments on replies, and maybe also  better data from future x-ray satellite missions. The significance of the detection depends, more than we'd wish, on dirty astrophysics involved in modeling the standard x-ray emission from galactic plasma. It seems unlikely that the sterile neutrino model with the originally reported parameters will stand, as it is in tension with several other analyses. The probability of the 3.5 keV signal being of dark matter origin is certainly much lower than a few months ago. But the jury is still out, and it's not impossible to imagine that more data and more analyses will tip the scales the other way.

Further reading: how to protect yourself from someone attacking you with a banana.


Andrew said...

It's worth noting that there are possible dark matter alternatives to decaying DM for this signal. Specifically, "exciting dark matter" (XDM) with nearly degenerate states that produce x-rays via collisional excitation and (relatively) rapid decay is interesting from this point of view. Since the lower velocity dispersions in dSph vs larger galaxies and galaxies vs clusters, XDM helps make sense of the pattern of constraints/detections if you assume a DM origin (see 1410.7766; disclaimer: I am an author). What we know about morphology of the signal in the Perseus cluster is confusing from the DM point of view (to say the least), so there may be some kind of atomic background at least there. Anyway, it's still an interesting area.

Joseph Conlon said...

You mean Andromeda rather than the Milky Way centre in your second bullet point.

Jester said...

Yes, thanks, corrected.

Theo Nieuwenhuizen said...

So mrs Bulbula gave birth to a still born Bulbulon, surely a drama, but that's family affairs. What concerns us is that for their decay Bulbulons (alas, also "Bulbula" in latin) need an extra Higgs sector; I thought we went over that already: Not now, please. Probably Bulbula are DNA-related to WIMPs: Would they exist, then there could not be a Galaxy full of Bulbulon/WIMP seekers.

Xezlec said...

I found today's plot more accessible than most.

Also, wouldn't the group opposing the bulbulon more accurately be described as a "bunch" rather than a "group"?

Theo Nieuwenhuizen said...


I also find these ongoing scientific disputes a waste of time. We should better decide democratically or appoint an emperor. Either way, we'll secure LCDM.

Anonymous said...

I think that you have accurately represented the current status of the debate.
There is an unknown signal at 3.5 keV in some (but not all galaxies.) The emission lines is fairly broad, so it rules out any single atomic transition line, but it doesn't rule out a grouping of lines, perhaps from potassium.
I think that people (myself included) jumped on the news of this emission line because it fits nicely with the idea that dark matter is a keV mass particle that can decay into light active neutrinos.
In my most recent post, I discuss a new paper by Salvatelli et al. who find experimental evidence for the theory that dark matter can decay into dark energy.

If keV-sterile-neutrinos could decay into meV-active-neutrinos (whose thermal and/or degeneracy pressure effectively acts like dark energy to expand the universe), then there is a consistent unification between multiple data sources. Dark matter = sterile neutrinos. & Dark Energy = pressure exerted by light active neutrinos.

One of the many questions that remains for me is: why are sterile neutrinos heavier than active neutrinos? This seems to contradict the trend that the higher the rest mass of a fermion, the more modes of interaction available to it. (neutrinos < electrons < quarks)

We really need a theory that can explain the rest mass of fundamental particles.

Xezlec said...


Waste of time? What are you talking about?

Theo Nieuwenhuizen said...

I did not appreciate to call one group of critics a "bunch". With such charged terms we seem not to desire open scientific discussion.

Xezlec said...

"Bunch" is a charged term?! It's just a more casual English synonym for "group". I was making a pun, something I probably shouldn't do on a blog frequented by non-native English speakers. Mea culpa I guess.

Theo Nieuwenhuizen said...

The experimental situation is very accurately represented in Figure 1 of this post. It shows the results of two groups, I would say, less the result of a group and a bunch.

Stefano Profumo said...

Dear Adam: you give a rather accurate summary of the situation. Two small corrections:

(1) in 1411.1759 we also fit the Andromeda X-ray flux in a broader energy range than 3-4 keV (specifically 3-7 keV), still not finding any evidence for a 3.5 keV line; we also explain why including the 2-3 keV range makes the fit problematic and poor enough that a line can be added at 3.5 keV (in addition to other lines); we also explain that the detected "lines" in the extended energy range are very likely unphysical, stemming from the poor modeling of the continuum.

(2) the Bulbul et al predictions for K XVIII are not based on same-element ratios, but on their multi-temperature models and on the brightness of other elemental lines (S XVI, Ca XIX and XX). Same-element ratios are however important checks, as they don't depend on relative elemental abundances. What we showed was that the multi-temperature models used in Bulbul et al are often biased towards large temperatures, leading to highly suppressed K XVIII predictions, making such predictions not conservative enough.


StevieB said...

Xezlec: never apologize for having a sense of humor to those who have none. Especially on this blog where wit is valued

Yohan said...

Hi Jester,
I just found articles on 2 science news and general websites speaking about the possible discovery of DM fro X-ray. It points to this article: What is strange is that this article is from February 2014 ... I'm not sure why it is surfacing only now ! Do you know about this one ? You don't seem to talk about it in this summary of 3,5 keV line.

Jester said...

1402.4119 is a paper that appeared a few days after Bulbul et al, and it contains claims of the 3.5 keV line emission from the Perseus cluster and the Andromeda galaxy. The 2nd bullet in my post concerns the dispute about the latter claim. I suppose the paper was recently published in some journal, and that's how it caught the eye of someone in some news outlet.

Kevork Abazajian said...

There are only a few analyses that place limits on the decay parameter space---which is commensurate with a flux limit. Those that do are plotted at my post here. For example, the Anderson+ stacked galaxy limit makes strong sensitivity claims, but is overwhelmed by systematics (their Fig. 4) so they do not have and cannot place a limit. The Horiuchi+ 2013 limit remains the most sensitive, including when compared to the Malyshev+ dwarf stack limit.