The Kes 41 point source is better described as a cloud
A reanalysis of more than 16 years of Fermi LAT data reports three extended gamma-ray components, including one at the Kes 41 supernova remnant that is statistically preferred over the previously reported point source.
A gamma-ray source previously reported as a point source is better described as an extended cloud, and it is not sitting where that report put it. The emission lines up with the natal cloud of the Kes 41 supernova remnant rather than with the remnant's shell, where the point source had been reported. In the same reanalysis, two more extended components appear toward the Westerlund 1 massive star cluster.
Nothing new was launched to find this. The authors took more than 16 years of Fermi LAT data above 0.8 GeV and asked whether clusters of catalogued gamma-ray sources with no counterpart at other wavelengths are better described by extended emission components than by point-like sources. The analysis reports three extended components. For Kes 41, the extended component is statistically preferred over the previously reported point source; the two near Westerlund 1 are reported as extended emission overlapping neutral gas. All three have soft spectra, meaning the emission falls off steeply with energy, and all three sit toward regions of high gas column density in the Galactic plane.
A point-source description gives the emission a position and a spectrum. An extended description gives it a size as well. That extra freedom is why the question has to be settled statistically, by fitting both models to the same photons and asking which the data prefer, rather than by eye.
Westerlund 1 is why the second half of that matters. There is growing evidence that massive stellar clusters accelerate cosmic rays. The two components near Westerlund 1 overlap neutral gas sitting up to about 100 pc, or 330 light-years, from the edge of the cluster's superbubble, the cavity its winds have carved in the surrounding material. That puts them at the cluster's outskirts rather than on top of it.
| Kes 41 | Near Westerlund 1 | |
|---|---|---|
| Previously reported as | a point source toward the supernova remnant shell | a cluster of gamma-ray sources with no multiwavelength match |
| Preferred description | one extended component, associated with the remnant's natal cloud | two extended components |
| Gas it overlaps | the remnant's own cloud | neutral gas within about 100 pc of the superbubble's edge |
| Spectrum | soft | soft |
| Fraction required, if the emission is local particle injection | at most 5% of the supernova remnant's energy | 10⁻⁴ of the cluster wind's mechanical power |
If the emission comes from accelerated particles rather than from gas, Kes 41 needs at most 5% of the supernova remnant's energy converted into cosmic rays, while the two components near Westerlund 1 need 10⁻⁴ of the cluster wind's mechanical power turned into gamma rays. Put both as plain fractions and the gap is easy to see: 5×10⁻² against 1×10⁻⁴, a difference of about 500. One ten-thousandth, in the units most people find easier, is 0.01%.
Those are fractions of two different budgets, the supernova remnant's energy and the cluster wind's mechanical power, so they are not the same quantity and should not be read as a ranking. The factor of 500 is my own arithmetic on the paper's two numbers; the paper does not make that comparison. What the fractions do show is how hard each explanation has to work. Whether the gap reflects genuinely different physics around a supernova remnant and a stellar cluster, or a sign that the extended emission is gas rather than particles, the paper leaves open, and that fork applies to all three components rather than to one of them.
The paper's own statement about the middle ground is narrow: particle transport in the vicinity of these sources appears to be influenced by processes markedly different from large-scale diffusion in the Milky Way. Here is the picture I take from that, and it is mine rather than the paper's. Diffusive transport over galactic distances erases the shape of any one accelerator, so structure at a scale larger than a single object but smaller than the Galactic background is where that erasure should first become visible.
Unassociated sources are the ones with no counterpart at another wavelength. If some of them are not objects at all but extended structures, that could account for part of the soft unassociated Galactic sources the Fermi LAT detects.
Two details make the new components suspicious in a useful way. They have soft spectra, and they sit on high gas column density. That is also what mismodeled gas in the interstellar background model would look like, and the paper offers that as one of two possible explanations. If the gas model is wrong in these directions, mismodeled gas will look like soft extended emission whether or not a particle was ever accelerated there. The authors keep both readings open, which makes the conversion efficiencies above conditional. No multiwavelength confirmation is reported.
"Statistically preferred" is a statement about which model fits the same photons better, not a photograph of a cloud. The Kes 41 change is a swap of which object gets the credit, the shell where the point source sat or the cloud the progenitor star formed in, and the fit prefers the cloud.
What would settle it is a better gas model along those sightlines, good enough to remove the gas explanation. Until then, three new extended components is what the analysis reports. Three new particle accelerators is a hypothesis about it. This is a preprint, not a settled result.