2007ApJ...660.1215T

Far-infrared Spitzer observations of elliptical galaxies are inconsistent with simple steady state models of dust creation in red giant stars and destruction by grain sputtering in the hot interstellar gas at T∼10⁷ K. The flux at 24 µm correlates with optical fluxes, suggesting that this relatively hot dust is largely circumstellar. But fluxes at 70 and 160 µm do not correlate with optical fluxes. Elliptical galaxies with similar L_B_ have luminosities at 70 and 160 µm (L₇₀and L₁₆₀) that vary over a factor of ∼100, implying an additional source of dust unrelated to that produced by ongoing local stellar mass loss. Neither L₇₀/L_B nor L₁₆₀/L_B correlate with the stellar age or metallicity. Optical line fluxes from warm gas at T∼10⁴ K correlate weakly with L₇₀ and L₁₆₀, suggesting that the dust may be responsible for cooling this gas. Many normal elliptical galaxies have emission at 70 µm that is extended to 5-10 kpc. Extended far-infrared emission with sputtering lifetimes of ∼10⁸ yr is difficult to maintain by mergers with gas-rich galaxies. Instead, we propose that this cold dust is buoyantly transported from reservoirs of dust in the galactic cores, which are supplied by mass loss from stars in the core. Intermittent energy outbursts from AGNs can drive the buoyant outflow.