Mysterious Gamma-Ray Excess Traces Stars

Using a newly developed tool, SkyFACT, which is a hybrid approach between template fitting and image reconstruction, we study the gamma-ray emission from the Galactic Center. We find that the morphology of the so-called “Fermi-LAT Galactic Center Excess” is better described by the distribution of stars in the boxy bulge rather than the expected dark matter density squared. This supports the interpretation that this emission is due to millisecond pulsars instead of dark matter.

Like Comment

In 2009 a signal in excess over the known backgrounds was detected in the direction of the Galactic Center in the data of the Fermi Large Area Telescope (LAT). Now, almost 10 years later, this signal has established itself as a persistent feature in the data, but its nature remains unclear. It was long believed that this so-called “Fermi Galactic Center Excess” - GCE for short - could be caused by annihilating dark matter particles. It has indeed all the features expected from a typical weakly-interacting massive particle (WIMP) dark matter candidate: the right energy spectrum, peaked at around few GeV as expected for WIMP masses of a few tens of GeV, and the right morphology, symmetric about the Galactic center, peaked towards that direction and extended up to high latitudes. The spatial distribution of the excess follows r-2.5 as a function of Galactocentric radius, r, and appears spherically symmetric. This is reminiscent of an annihilating dark matter particle distributed according to the Navarro-Frenk-White (NFW) profile often suggested by numerical simulations of structure formation.

However, it was pointed out that old and fast spinning neutron stars called millisecond pulsars (MSPs) could also be responsible for the signal. Their gamma-ray emission has a spectrum very similar to that of the excess. Moreover, two complementary studies in 2015, one by a group from the United States and the other by our group in Amsterdam, found corroborative evidence that the statistics of the gamma-ray data actually is more readily explained by a distribution of discrete sources, such as stars, rather than something that is smoothly distributed, such as dark matter. Nevertheless, the question of why these stars would be distributed in such a similar way to dark matter remained open.

Most previous analyses studied the GCE using a technique called template fitting. In these studies a set of spatial templates, representing different emission components, is fit to the data with a free normalization in each energy bin. The template for the GCE that worked best was an NFW template. However, this technique has obvious pitfalls, most importantly it cannot account for spatial uncertainties and errors in the various spatial templates. Last year, my collaborators developed a new code, called SkyFACT, which is a hybrid approach between template fitting and image reconstruction, i.e. it fits spatial templates, but the templates also have freedom to adapt themselves spatially to the data. An obvious use case for this tool is to re-evaluate the characteristics of the GCE, and this is what we set out to do.

(a) What the spatial distribution of light from dark matter annihilation in the inner Galaxy would look like. (b) The distribution of the light from a population of MSPs in the Galactic boxy- and nuclear bulge.

In our analysis of the GCE with SkyFACT, we used not only dark-matter-inspired templates but also templates that trace the distribution of stellar mass in the inner Galaxy. In particular, a characteristic feature of the Milky Way, other than its spiral arms, is the central Bulge, which accounts for about 15% of the total stellar mass. Interestingly, the distribution of light expected from a population of stars in the Galactic Bulge is not very different from that of dark matter! However, the bulge is box shaped and therefore the emission would not be completely spherically symmetric. Quite surprisingly, the explicit spatial profile of the Bulge had not been used before to study the GCE. However, previous studies did find that spherical symmetry was preferred for the GCE, rather than some more oblate profile. Instead, with our new analysis tool we found the stellar distribution provides a better fit to the data than dark matter.

The correlation between gamma-ray (>100 MeV) emission and stellar mass. The GCE excess emission is absorbed by the boxy-bulge (blue) and near the Galactic Centre by the nuclear bulge (green). Gamma-ray emission per unit-stellar-mass is consistent between these two components (black-dashed line is constant emission per unit stellar mass).

This left us with the task of reconciling our new results with those of previous analyses. For this purpose we performed an extensive effort to bridge the gap between our analysis and the traditional template fitting analyses, by removing the additional freedom that SkyFACT offers one step at a time. This way we found our results to be consistent with earlier analyses, and our new findings to be a consequence of our ability to better model fore- and backgrounds with SkyFACT.

10 years after the initial discovery of the GCE we are in the position that the dark matter interpretation is slowly losing support. There are at this time two pieces of corroborative evidence for the MSP interpretation. First, analyses of photon statistics suggested that the GCE was composed of discrete sources. Second, we now find using SkyFACT that the morphology appears to trace stellar mass rather than dark matter. Another recent study published in Nature Astronomy by an independent group using the traditional template fitting technique, but with with improved fore- and background templates, supports this conclusion. Nevertheless, in order to fully settle the GCE mystery in favour of MSPs, we have to detect them directly. This will require detection of their pulsed signals, which, if they exist, is likely to happen in the next few years with upcoming radio telescopes!

Our paper can be found here.

Richard Bartels

PhD student, University of Amsterdam