The hot gas atmospheres of massive elliptical galaxies should cool. The gas radiates X-rays, loses energy, and should condense into cold clouds that form stars. This is the cooling flow problem — the expected condensation exceeds the observed star formation by an order of magnitude. Something heats the gas, preventing it from cooling. The leading candidate is AGN feedback: jets from the central supermassive black hole inflate bubbles of relativistic plasma that displace the X-ray gas, creating cavities visible as depressions in the X-ray surface brightness.
Temi, Ubertosi, Brighenti, and colleagues (arXiv 2602.22415, February 2026) find that the AGN doesn't just prevent cooling — it creates it, but in specific locations. Multi-wavelength observations using MUSE (ionized gas), ALMA (molecular gas), and infrared (dust) reveal that cold, dusty gas clouds cluster preferentially around the rims of X-ray cavities. Not in the center of the cooling flow. Not randomly distributed through the hot atmosphere. At the edges of the bubbles that the AGN inflated.
The mechanism is compressive. The expanding AGN bubble pushes against the surrounding hot gas, creating a shell of enhanced density at the cavity rim. The compressed gas cools faster — radiative cooling rates scale with the square of the density. What would take billions of years to cool in the undisturbed atmosphere cools in millions of years in the compressed rim. The AGN simultaneously heats the atmosphere globally (preventing the bulk cooling flow) and cools it locally (triggering condensation at bubble edges).
The dust properties vary systematically. Where dust, molecular gas, and ionized gas coexist — the mixed-phase zones at cavity rims — the dust grains are smaller or less processed, with shallower extinction curves. Isolated dust clouds elsewhere in the galaxy have different grain properties: larger, more processed. The environmental conditions write their history into the dust. Rapid formation in the violent rim environment produces different grains than slow accumulation in quiescent regions.
The feedback is selective. The AGN solves the cooling flow problem by redistributing where cooling happens, not by eliminating cooling entirely. The destruction of the global flow creates local flows at the cavity edges. Heating begets cooling, but only at the boundary.